Background The application of deep learning in Alzheimer's disease (AD) diagnosis has shown promise, but most studies focus on Western populations, potentially limiting their applicability in African contexts. There is a critical need for validated diagnostic tools that account for population‐specific characteristics in neuroimaging analysis. Method We developed a transfer learning‐enhanced DenseNet121 architecture for AD classification. The model was initially pre‐trained on the OASIS dataset to learn general AD‐related features, followed by fine‐tuning on a local dataset from the University College Hospital (UCH), Ibadan, Nigeria. The local dataset comprised 140 subjects (63 dementia, 77 non‐dementia cases). Advanced preprocessing techniques, including skull‐stripping, spatial normalization, and grey matter segmentation, were applied to optimize image quality and feature extraction. Result Our model achieved exceptional performance metrics with an accuracy of 97.32% and an AUC score of 0.9916. The sensitivity and specificity were 98.37% and 96.04% respectively, with a precision of 96.80% and an F1 score of 97.58%. This performance significantly surpasses previous studies and demonstrates the effectiveness of our transfer learning approach in capturing population‐specific characteristics while maintaining high diagnostic accuracy. Conclusion The successful development and validation of our population‐specific model represents a significant advancement in AD diagnosis for African populations. The high performance metrics validate our transfer learning approach and demonstrate that high‐quality AD diagnosis models can be developed for specific populations while leveraging existing datasets for initial feature learning. This work provides a framework for developing locally‐validated diagnostic tools in low‐resource settings.
{"title":"Enhanced Deep Learning Model for Alzheimer's Disease Classification Using Brain MRI: A Nigerian Population Study","authors":"Ayokunle Joshua Ola","doi":"10.1002/alz70856_107749","DOIUrl":"https://doi.org/10.1002/alz70856_107749","url":null,"abstract":"Background The application of deep learning in Alzheimer's disease (AD) diagnosis has shown promise, but most studies focus on Western populations, potentially limiting their applicability in African contexts. There is a critical need for validated diagnostic tools that account for population‐specific characteristics in neuroimaging analysis. Method We developed a transfer learning‐enhanced DenseNet121 architecture for AD classification. The model was initially pre‐trained on the OASIS dataset to learn general AD‐related features, followed by fine‐tuning on a local dataset from the University College Hospital (UCH), Ibadan, Nigeria. The local dataset comprised 140 subjects (63 dementia, 77 non‐dementia cases). Advanced preprocessing techniques, including skull‐stripping, spatial normalization, and grey matter segmentation, were applied to optimize image quality and feature extraction. Result Our model achieved exceptional performance metrics with an accuracy of 97.32% and an AUC score of 0.9916. The sensitivity and specificity were 98.37% and 96.04% respectively, with a precision of 96.80% and an F1 score of 97.58%. This performance significantly surpasses previous studies and demonstrates the effectiveness of our transfer learning approach in capturing population‐specific characteristics while maintaining high diagnostic accuracy. Conclusion The successful development and validation of our population‐specific model represents a significant advancement in AD diagnosis for African populations. The high performance metrics validate our transfer learning approach and demonstrate that high‐quality AD diagnosis models can be developed for specific populations while leveraging existing datasets for initial feature learning. This work provides a framework for developing locally‐validated diagnostic tools in low‐resource settings.","PeriodicalId":7471,"journal":{"name":"Alzheimer's & Dementia","volume":"3 1","pages":""},"PeriodicalIF":14.0,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145938070","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Gustavo E Juantorena, Waleska Berrios, Maria Cecilia Fernández, Agustin Ibanez, Agustin Petroni, Juan E Kamienkowski
Background We extended our computerized Trail Making Test (c‐TMT) to investigate deficits in Mild Cognitive Impairment (MCI) compared to neurotypical controls. By integrating hand and eye tracking, we captured fine‐grained movement dynamics, revealing distinct trajectory alterations in MCI patients. These differences suggest potential digital biomarkers, offering a more precise assessment beyond traditional total time measurements. Methods Twenty‐nine MCI patients and 28 age‐ and education‐matched controls (with significant Mini‐Mental Test differences, p < 0.001) were enrolled at Hospital Italiano de Buenos Aires, Argentina, with informed consent. Two practice trials and 20 experimental trials (alternating TMT‐A and TMT‐B) were presented. Stimuli were displayed on a 24‐inch screen. Gaze was recorded from the right eye at 500 Hz using an EyeLink 1000 Plus. The mouse trajectory was displayed in real‐time, with feedback on the correct element selection. Results Linear Mixed Models (LMM) were applied to correct trials to estimate the main effects of subject group (MCI vs. control), trial type (TMT‐A vs. TMT‐B), and their interaction using the statsmodels library in Python. For performance metrics, LMM revealed a significant effect of subject group and trial type on the percentage of completion (PC) (SE = 0.066, p = 0.040; SE = ‐9.017, p = 1.9 × 10 −19 ) and the time required to complete a trial (RT) (SE = ‐2.514, p = 0.012; SE = 7.896, p = 2.9 × 10 −15 ). For eye‐tracking metrics, we found significant differences for both trial type (SE = 2.06, p = 0.002) and subject group (SE = 2.81, p = 0.023) in scanpath length (number of fixations). However, fixation duration differences were not significant (SE = 7.830, p = 0.68; SE = 12.90, p = 0.80). We also analyzed eye‐hand coordination by parsing fixations based on mouse position and time‐locking mouse and hand movements to target entry. Differences were observed by trial type but not by subject group. Conclusions Our c‐TMT version identified significant differences in scanpath length between MCI patients and controls. Hand and eye movements together allow fixation analysis to determine how increased fixations are distributed. These findings highlight the potential of this approach in Digital Neuropsychology.
背景：我们扩展了计算机跟踪测试（c - TMT），以研究轻度认知障碍（MCI）患者与神经正常对照组相比的缺陷。通过整合手和眼动追踪，我们捕捉到了细粒度的运动动态，揭示了MCI患者明显的轨迹改变。这些差异暗示了潜在的数字生物标志物，提供了比传统的总时间测量更精确的评估。方法29例MCI患者和28例年龄和教育程度相匹配的对照组（具有显著的Mini - Mental Test差异，p < 0.001）在阿根廷布宜诺斯艾利斯意大利医院登记，并获得知情同意。提出了两个实践试验和20个实验试验（TMT‐A和TMT‐B交替进行）。刺激被显示在24英寸的屏幕上。使用EyeLink 1000 Plus以500赫兹的频率记录右眼的注视。鼠标轨迹实时显示，并对正确的元素选择进行反馈。结果线性混合模型（LMM）应用于校正试验，以估计受试者组（MCI vs. control）、试验类型（TMT‐A vs. TMT‐B）的主要效应，并使用Python中的statmodels库进行交互。对于绩效指标，LMM显示受试者组和试验类型对完成百分比（PC） （SE = 0.066, p = 0.040; SE =‐9.017,p = 1.9 × 10−19）和完成试验所需时间（RT） （SE =‐2.514,p = 0.012; SE = 7.896, p = 2.9 × 10−15）有显著影响。对于眼动追踪指标，我们发现两种试验类型（SE = 2.06, p = 0.002）和受试者组（SE = 2.81, p = 0.023）在扫描路径长度（注视次数）上存在显著差异。注视时间差异无统计学意义（SE = 7.830, p = 0.68; SE = 12.90, p = 0.80）。我们还通过分析基于鼠标位置和锁定时间的鼠标和手的运动来分析眼手协调。不同试验类型观察到差异，但受试者组之间没有差异。我们的c - TMT版本确定了MCI患者和对照组之间扫描路径长度的显著差异。手和眼的运动一起允许注视分析来确定增加的注视是如何分布的。这些发现突出了这种方法在数字神经心理学中的潜力。
{"title":"Enhancing MCI Assessment: A Digital Trail Making Test with Integrated Eye and Hand Tracking","authors":"Gustavo E Juantorena, Waleska Berrios, Maria Cecilia Fernández, Agustin Ibanez, Agustin Petroni, Juan E Kamienkowski","doi":"10.1002/alz70856_107204","DOIUrl":"https://doi.org/10.1002/alz70856_107204","url":null,"abstract":"Background We extended our computerized Trail Making Test (c‐TMT) to investigate deficits in Mild Cognitive Impairment (MCI) compared to neurotypical controls. By integrating hand and eye tracking, we captured fine‐grained movement dynamics, revealing distinct trajectory alterations in MCI patients. These differences suggest potential digital biomarkers, offering a more precise assessment beyond traditional total time measurements. Methods Twenty‐nine MCI patients and 28 age‐ and education‐matched controls (with significant Mini‐Mental Test differences, <jats:italic>p</jats:italic> < 0.001) were enrolled at Hospital Italiano de Buenos Aires, Argentina, with informed consent. Two practice trials and 20 experimental trials (alternating TMT‐A and TMT‐B) were presented. Stimuli were displayed on a 24‐inch screen. Gaze was recorded from the right eye at 500 Hz using an EyeLink 1000 Plus. The mouse trajectory was displayed in real‐time, with feedback on the correct element selection. Results Linear Mixed Models (LMM) were applied to correct trials to estimate the main effects of subject group (MCI vs. control), trial type (TMT‐A vs. TMT‐B), and their interaction using the statsmodels library in Python. For performance metrics, LMM revealed a significant effect of subject group and trial type on the percentage of completion (PC) (SE = 0.066, <jats:italic>p</jats:italic> = 0.040; SE = ‐9.017, <jats:italic>p</jats:italic> = 1.9 × 10 <jats:sup>−19</jats:sup> ) and the time required to complete a trial (RT) (SE = ‐2.514, <jats:italic>p</jats:italic> = 0.012; SE = 7.896, <jats:italic>p</jats:italic> = 2.9 × 10 <jats:sup>−15</jats:sup> ). For eye‐tracking metrics, we found significant differences for both trial type (SE = 2.06, <jats:italic>p</jats:italic> = 0.002) and subject group (SE = 2.81, <jats:italic>p</jats:italic> = 0.023) in scanpath length (number of fixations). However, fixation duration differences were not significant (SE = 7.830, <jats:italic>p</jats:italic> = 0.68; SE = 12.90, <jats:italic>p</jats:italic> = 0.80). We also analyzed eye‐hand coordination by parsing fixations based on mouse position and time‐locking mouse and hand movements to target entry. Differences were observed by trial type but not by subject group. Conclusions Our c‐TMT version identified significant differences in scanpath length between MCI patients and controls. Hand and eye movements together allow fixation analysis to determine how increased fixations are distributed. These findings highlight the potential of this approach in Digital Neuropsychology.","PeriodicalId":7471,"journal":{"name":"Alzheimer's & Dementia","volume":"46 1","pages":""},"PeriodicalIF":14.0,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145947365","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Vidya Somashekarappa, Meera Srikrishna, Silke Kern, Joyce R Chong, Eric Westman, Christopher Chen, Ingmar Skoog, Jakob Seidlitz, Michael Schöll
Background Brain tissue segmentation is vital in Alzheimer's and dementia research for creating detailed neuroanatomical maps, diagnosing early‐stage neurodegeneration, and guiding interventions. Although MRI remains the standard approach for its superior soft‐tissue contrast, CT is a more accessible imaging modality in acute and resource‐constrained settings. Method This study utilized paired CT‐MRI datasets from the Gothenburg H70 Birth Cohort ( N = 733) and the Memory Clinic Cohort of the National University Hospital, Singapore (NUS Dementia Cohort, N = 210) to train and evaluate advanced segmentation models— nnUNet (2D & 3D models for 300‐1000 epochs) and MedNeXt (3D‐ Small, Base, Medium and Large models for 3x3x3 & 5x5x5 kernels). MRI‐derived labels were employed to guide CT segmentation, allowing accurate delineation of brain tissue segmentation (Gray Matter: GM, White Matter: WM and Cerebrospinal Fluid: CSF). Evaluation was conducted on all axial datasets for all variations of the models and for coronal & sagittal orientations the best performing models were utilized for inference. Result The 3D nnU‐Net achieved average Dice Similarity Coefficients (DSCs) of 0.82, 0.72, and 0.76 for axial, coronal, and sagittal orientations, respectively, while MedNeXt demonstrated slightly superior performance with DSCs of 0.83, 0.73, and 0.78. MedNeXt also exhibited improved volumetric similarity in axial datasets, with scores ranging from 0.842 (CSF, sagittal) to 0.992 (WM, axial). When applied to dementia cohorts, MedNeXt achieved higher generalizability with an average DSC and volumetric similarity of 0.73 and 0.912, compared to 0.70 and 0.854 for nnU‐Net. Extended training (1000 epochs) enhanced nnU‐Net's performance, yet MedNeXt displayed superior scalability, handling larger kernel sizes and multi‐modal imaging scenarios. However, significantly longer training times of up to 288 hours was required for the largest model. Conclusion Automated CT brain segmentation guided by MRI‐derived labels demonstrates clinically acceptable segmentation performance on untrained dementia cohort. nnU‐Net is more resource‐efficient and suitable for limited‐resource settings, while MedNeXt has higher accuracy excelling in multi‐orientation and multi‐modal datasets. These findings validate the feasibility of using CT imaging with advanced segmentation frameworks to develop accessible neuroimaging tools for Alzheimer's and dementia research, addressing diagnostic challenges across diverse clinical contexts.
{"title":"Automated Brain Tissue Segmentation on CT guided by MRI: Advancing AI‐based Neuroimaging for Dementia","authors":"Vidya Somashekarappa, Meera Srikrishna, Silke Kern, Joyce R Chong, Eric Westman, Christopher Chen, Ingmar Skoog, Jakob Seidlitz, Michael Schöll","doi":"10.1002/alz70856_104843","DOIUrl":"https://doi.org/10.1002/alz70856_104843","url":null,"abstract":"Background Brain tissue segmentation is vital in Alzheimer's and dementia research for creating detailed neuroanatomical maps, diagnosing early‐stage neurodegeneration, and guiding interventions. Although MRI remains the standard approach for its superior soft‐tissue contrast, CT is a more accessible imaging modality in acute and resource‐constrained settings. Method This study utilized paired CT‐MRI datasets from the Gothenburg H70 Birth Cohort ( <jats:italic>N</jats:italic> = 733) and the Memory Clinic Cohort of the National University Hospital, Singapore (NUS Dementia Cohort, <jats:italic>N</jats:italic> = 210) to train and evaluate advanced segmentation models— nnUNet (2D & 3D models for 300‐1000 epochs) and MedNeXt (3D‐ Small, Base, Medium and Large models for 3x3x3 & 5x5x5 kernels). MRI‐derived labels were employed to guide CT segmentation, allowing accurate delineation of brain tissue segmentation (Gray Matter: GM, White Matter: WM and Cerebrospinal Fluid: CSF). Evaluation was conducted on all axial datasets for all variations of the models and for coronal & sagittal orientations the best performing models were utilized for inference. Result The 3D nnU‐Net achieved average Dice Similarity Coefficients (DSCs) of 0.82, 0.72, and 0.76 for axial, coronal, and sagittal orientations, respectively, while MedNeXt demonstrated slightly superior performance with DSCs of 0.83, 0.73, and 0.78. MedNeXt also exhibited improved volumetric similarity in axial datasets, with scores ranging from 0.842 (CSF, sagittal) to 0.992 (WM, axial). When applied to dementia cohorts, MedNeXt achieved higher generalizability with an average DSC and volumetric similarity of 0.73 and 0.912, compared to 0.70 and 0.854 for nnU‐Net. Extended training (1000 epochs) enhanced nnU‐Net's performance, yet MedNeXt displayed superior scalability, handling larger kernel sizes and multi‐modal imaging scenarios. However, significantly longer training times of up to 288 hours was required for the largest model. Conclusion Automated CT brain segmentation guided by MRI‐derived labels demonstrates clinically acceptable segmentation performance on untrained dementia cohort. nnU‐Net is more resource‐efficient and suitable for limited‐resource settings, while MedNeXt has higher accuracy excelling in multi‐orientation and multi‐modal datasets. These findings validate the feasibility of using CT imaging with advanced segmentation frameworks to develop accessible neuroimaging tools for Alzheimer's and dementia research, addressing diagnostic challenges across diverse clinical contexts.","PeriodicalId":7471,"journal":{"name":"Alzheimer's & Dementia","volume":"185 1","pages":""},"PeriodicalIF":14.0,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145947384","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Brahyan J Galindo Mendez, Ling Teng, Gad A. Marshall, Lei Liu, Jasmeer P. Chhatwal, Timothy J. Hohman, Richard Mayeux, Philip L. De Jager, Robert A. Rissman, Keith A. Johnson, Reisa A. Sperling, Hyun‐Sik Yang
Background Alzheimer's disease (AD) is a highly heritable neurodegenerative disorder, and human genetics have strongly implicated microglia (Mic) in AD pathogenesis. Leveraging our novel method to derive cell‐type‐specific AD polygenic risk scores (cts‐ADPRS), we examined their association with longitudinal plasma‐phospho‐tau‐217 (pTau217) and cognition in cognitively unimpaired (CU) older adults. Methods We analyzed longitudinal data from a secondary AD prevention trial (A4; CU with elevated+Aβ and LEARN; CU with sub‐threshold ‐Aβ). cts‐ADPRS were derived and standardized using prior published method by (1) excluding APOE and selecting top 10% of genes specifically expressed in each major brain cell type (excitatory neurons, inhibitory neurons, astrocyte, microglia [Mic], oligodendrocyte, oligodendrocyte progenitor cells) and (2) deriving each cts‐ADPRS using the variants within these genes ± 30 kb. We analyzed the relationship of each cts‐ADPRS with longitudinal change in pTau217 and Preclinical Alzheimer Cognitive Composite (PACC). We used linear mixed effect models (LMEM) adjusted for baseline Aβ PET (florbetapir) cortical composite, age, sex, APOE ε4/ε2, years of education, first three genotype principal components, and their time interaction terms. We extracted adjusted random slopes of pTau217 and PACC from LMEM and, performed a mediation analysis to examine the relationship among cts‐ADPRS, pTau217, and PACC. Results We included 1179 CU subjects ( n = 474 females (40%), 70.9 ± 4.5 years old) of European descent who had pTau217 and PACC. Mic‐ADPRS was associated with a longitudinal increase in pTau217 (beta = 13.9 x, SE = 2.4 x, p = <0.001) while none of the other cts‐ADPRS were (all p >0.05). Mic‐ADPRS was also associated with faster PACC decline (beta= ‐15.0 x, SE = 1.8 x, p = < 0.001). Mediation analysis suggests 42% of Mic‐ADPRS–PACC association is mediated by increased pTau217. Conclusion Our findings suggest AD heritability localizing to microglial genes and contribute to increased soluble pTau217 release at a given Aβ burden, which in turn mediates the association between microglial AD genetic risk and cognitive decline. Our results are consistent with previous studies implying microglia in Aβ‐related tau accumulation and indicate that much of microglial impact on cognitive decline may occur through accelerated tau pathology in CU older adults.
阿尔茨海默病（AD）是一种高度遗传性的神经退行性疾病，人类遗传学强烈地暗示了小胶质细胞（Mic）在AD发病机制中的作用。利用我们的新方法获得细胞类型特异性AD多基因风险评分（cts - ADPRS），我们研究了它们与纵向血浆磷酸化- tau - 217 （pTau217）和认知能力之间的关系。方法：我们分析了来自二级AD预防试验的纵向数据（A4； CU与+ a β和LEARN升高；CU与亚阈值- a β升高）。cts‐ADPRS的推导和标准化使用先前发表的方法：(1)排除APOE并选择在每种主要脑细胞类型（兴奋性神经元、抑制性神经元、星形胶质细胞、小胶质细胞[Mic]、少突胶质细胞、少突胶质细胞祖细胞）中特异性表达的前10%的基因；(2)使用这些基因内±30 kb的变异来推导每个cts‐ADPRS。我们分析了每个cts‐ADPRS与pTau217和临床前阿尔茨海默认知复合物（PACC）纵向变化的关系。我们使用线性混合效应模型（LMEM）调整基线Aβ PET （florbetapir）皮质复合物、年龄、性别、APOE ε4/ε2、受教育年限、前三个基因型主成分及其时间相互作用项。我们从LMEM中提取了调整后的pTau217和PACC的随机斜率，并进行了中介分析，以检验cts - ADPRS、pTau217和PACC之间的关系。结果我们纳入了1179名CU受试者（n = 474名女性（40%），70.9±4.5岁），欧洲血统，患有pTau217和PACC。Mic‐ADPRS与pTau217的纵向增加相关（beta = 13.9 x, SE = 2.4 x, p = <0.001），而其他cts‐ADPRS与此无关（均p >；0.05）。Mic‐ADPRS也与更快的PACC下降相关（beta= - 15.0 x, SE = 1.8 x, p = < 0.001）。中介分析表明，42%的Mic - ADPRS-PACC关联是由pTau217升高介导的。我们的研究结果表明，阿尔茨海默病的遗传性定位于小胶质基因，并有助于在给定的a β负荷下增加可溶性pTau217的释放，这反过来介导了小胶质阿尔茨海默病遗传风险与认知能力下降之间的关联。我们的研究结果与之前的研究一致，表明小胶质细胞参与了Aβ相关的tau积累，并表明在CU老年人中，小胶质细胞对认知能力下降的影响可能是通过加速tau病理发生的。
{"title":"Microglia‐specific Alzheimer's disease polygenic risk score predicts longitudinal increase in plasma tau and faster cognitive decline in cognitively unimpaired older adults","authors":"Brahyan J Galindo Mendez, Ling Teng, Gad A. Marshall, Lei Liu, Jasmeer P. Chhatwal, Timothy J. Hohman, Richard Mayeux, Philip L. De Jager, Robert A. Rissman, Keith A. Johnson, Reisa A. Sperling, Hyun‐Sik Yang","doi":"10.1002/alz70856_106441","DOIUrl":"https://doi.org/10.1002/alz70856_106441","url":null,"abstract":"Background Alzheimer's disease (AD) is a highly heritable neurodegenerative disorder, and human genetics have strongly implicated microglia (Mic) in AD pathogenesis. Leveraging our novel method to derive cell‐type‐specific AD polygenic risk scores (cts‐ADPRS), we examined their association with longitudinal plasma‐phospho‐tau‐217 (pTau217) and cognition in cognitively unimpaired (CU) older adults. Methods We analyzed longitudinal data from a secondary AD prevention trial (A4; CU with elevated+Aβ and LEARN; CU with sub‐threshold ‐Aβ). cts‐ADPRS were derived and standardized using prior published method by (1) excluding APOE and selecting top 10% of genes specifically expressed in each major brain cell type (excitatory neurons, inhibitory neurons, astrocyte, microglia [Mic], oligodendrocyte, oligodendrocyte progenitor cells) and (2) deriving each cts‐ADPRS using the variants within these genes ± 30 kb. We analyzed the relationship of each cts‐ADPRS with longitudinal change in pTau217 and Preclinical Alzheimer Cognitive Composite (PACC). We used linear mixed effect models (LMEM) adjusted for baseline Aβ PET (florbetapir) cortical composite, age, sex, <jats:italic>APOE</jats:italic> ε4/ε2, years of education, first three genotype principal components, and their time interaction terms. We extracted adjusted random slopes of pTau217 and PACC from LMEM and, performed a mediation analysis to examine the relationship among cts‐ADPRS, pTau217, and PACC. Results We included 1179 CU subjects ( <jats:italic>n</jats:italic> = 474 females (40%), 70.9 ± 4.5 years old) of European descent who had pTau217 and PACC. Mic‐ADPRS was associated with a longitudinal increase in pTau217 (beta = 13.9 x, SE = 2.4 x, <jats:italic>p</jats:italic> = <0.001) while none of the other cts‐ADPRS were (all <jats:italic>p</jats:italic> >0.05). Mic‐ADPRS was also associated with faster PACC decline (beta= ‐15.0 x, SE = 1.8 x, <jats:italic>p</jats:italic> = < 0.001). Mediation analysis suggests 42% of Mic‐ADPRS–PACC association is mediated by increased pTau217. Conclusion Our findings suggest AD heritability localizing to microglial genes and contribute to increased soluble pTau217 release at a given Aβ burden, which in turn mediates the association between microglial AD genetic risk and cognitive decline. Our results are consistent with previous studies implying microglia in Aβ‐related tau accumulation and indicate that much of microglial impact on cognitive decline may occur through accelerated tau pathology in CU older adults.","PeriodicalId":7471,"journal":{"name":"Alzheimer's & Dementia","volume":"253 1","pages":""},"PeriodicalIF":14.0,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145947387","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Extracellular vesicles (EVs) circulating in blood hold significant diagnostic potential for various diseases, including neurological disorders. However, as a living tissue, blood undergoes substantial changes during ex vivo transition and specimen handling. Pre-analytical factors, such as the choice of blood collection tubes and blood processing methods, can activate blood cells (e.g., platelets, neutrophils, monocytes), potentially altering EV composition. To assess the effects of these pre-analytical factors on the stability of EV composition in blood specimens, we examined EV populations in plasma samples from healthy donors following various collection and processing methods.
Method
MACSPlex EV Kits were used to profile multiple EV populations simultaneously in a single sample via flow cytometry. Using Immuno-Oncology and Neuro EV panels, we evaluated 63 EV populations in blood collected in three types of tubes: EDTA, ACD-A and Protein Plus BCT, processed at various times. We also paired these assays with three EV isolation methods, including precipitation and size exclusion chromatography, to assess the impact of EV enrichment on EV profiling.
Result
When plasma isolation was delayed, certain EV populations in blood increased significantly during storage in both EDTA and ACD-A tubes, particularly those labeled with surface markers related to immune response and inflammation (e.g., CD9, CD14, CD24, CD29, CD45) and activated platelets (e.g., CD42a, CD62P). Conversely, the overall EV profile in Protein Plus BCT remained close to the draw-time composition. EV isolation methods enhanced EV binding to capture beads, enabling more reliable measurement of circulating EVs associated with the nervous system (e.g., GLAST, PSA-NCAM, VGLUT2).
Conclusion
Our findings demonstrate that EV composition remains stable in Protein Plus BCT compared to EDTA and ACD-A tubes, with a small number of exceptions. This insight into EV stability in blood specimens is crucial for advancing EV biomarker research in neurological disorders, including Alzheimer's disease, potentially enhancing diagnostic accuracy and therapeutic strategies.
背景：血液循环中的细胞外囊泡（EVs）对包括神经系统疾病在内的多种疾病具有重要的诊断潜力。然而，作为一种活组织，血液在离体转化和标本处理过程中经历了实质性的变化。分析前因素，如采血管的选择和血液处理方法，可以激活血细胞（如血小板、中性粒细胞、单核细胞），潜在地改变EV组成。为了评估这些分析前因素对血液标本中EV组成稳定性的影响，我们通过不同的收集和处理方法检测了健康献血者血浆样本中的EV种群。方法采用MACSPlex EV试剂盒，通过流式细胞术同时对单个样品中的多个EV群体进行分析。使用免疫肿瘤学和神经细胞组，我们评估了三种不同时间处理的EDTA、ACD - A和Protein Plus BCT采集的血液中的63个细胞群。我们还将这些实验与三种EV分离方法（包括沉淀法和粒径排除色谱法）配对，以评估EV富集对EV谱分析的影响。当血浆分离延迟时，EDTA和ACD‐A管中储存的血液中某些EV数量显著增加，特别是那些标记有与免疫反应和炎症相关的表面标志物（如CD9、CD14、CD24、CD29、CD45）和活化血小板（如CD42a、CD62P）的EV数量。相反，Protein Plus BCT的整体EV谱仍然接近于绘制时间组成。EV分离方法增强了EV与捕获珠粒的结合，从而能够更可靠地测量与神经系统相关的循环EV（例如，GLAST, PSA - NCAM, VGLUT2）。我们的研究结果表明，与EDTA和ACD‐A管相比，在Protein Plus BCT中，EV成分保持稳定，只有少数例外。对血液标本中EV稳定性的了解对于推进包括阿尔茨海默病在内的神经系统疾病的EV生物标志物研究至关重要，有可能提高诊断准确性和治疗策略。
{"title":"Stability of Extracellular Vesicle Composition in Blood Specimens: Impact of Pre-analytical Factors and Isolation Methods","authors":"Eunju Seong","doi":"10.1002/alz70856_105924","DOIUrl":"10.1002/alz70856_105924","url":null,"abstract":"<div>\u0000 \u0000 \u0000 <section>\u0000 \u0000 <h3> Background</h3>\u0000 \u0000 <p>Extracellular vesicles (EVs) circulating in blood hold significant diagnostic potential for various diseases, including neurological disorders. However, as a living tissue, blood undergoes substantial changes during ex vivo transition and specimen handling. Pre-analytical factors, such as the choice of blood collection tubes and blood processing methods, can activate blood cells (e.g., platelets, neutrophils, monocytes), potentially altering EV composition. To assess the effects of these pre-analytical factors on the stability of EV composition in blood specimens, we examined EV populations in plasma samples from healthy donors following various collection and processing methods.</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Method</h3>\u0000 \u0000 <p>MACSPlex EV Kits were used to profile multiple EV populations simultaneously in a single sample via flow cytometry. Using Immuno-Oncology and Neuro EV panels, we evaluated 63 EV populations in blood collected in three types of tubes: EDTA, ACD-A and Protein Plus BCT, processed at various times. We also paired these assays with three EV isolation methods, including precipitation and size exclusion chromatography, to assess the impact of EV enrichment on EV profiling.</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Result</h3>\u0000 \u0000 <p>When plasma isolation was delayed, certain EV populations in blood increased significantly during storage in both EDTA and ACD-A tubes, particularly those labeled with surface markers related to immune response and inflammation (e.g., CD9, CD14, CD24, CD29, CD45) and activated platelets (e.g., CD42a, CD62P). Conversely, the overall EV profile in Protein Plus BCT remained close to the draw-time composition. EV isolation methods enhanced EV binding to capture beads, enabling more reliable measurement of circulating EVs associated with the nervous system (e.g., GLAST, PSA-NCAM, VGLUT2).</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Conclusion</h3>\u0000 \u0000 <p>Our findings demonstrate that EV composition remains stable in Protein Plus BCT compared to EDTA and ACD-A tubes, with a small number of exceptions. This insight into EV stability in blood specimens is crucial for advancing EV biomarker research in neurological disorders, including Alzheimer's disease, potentially enhancing diagnostic accuracy and therapeutic strategies.</p>\u0000 </section>\u0000 </div>","PeriodicalId":7471,"journal":{"name":"Alzheimer's & Dementia","volume":"21 S2","pages":""},"PeriodicalIF":11.1,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://alz-journals.onlinelibrary.wiley.com/doi/epdf/10.1002/alz70856_105924","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145947406","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Nelly Joseph-Mathurin, Katherine Gong, Gengsheng Chen, Parinaz Massoumzadeh, Jeremy F. Strain, Laura Ibanez, Brian A. Gordon, Jorge J. Llibre-Guerra, Jason J. Hassenstab, Richard J. Perrin, Chengjie Xiong, Randall J. Bateman, Eric McDade, Tammie L.S. Benzinger, Carlos Cruchaga, Dominantly Inherited Alzheimer Network
<div> <section> <h3> Background</h3> <p>With the advent of disease-modifying treatment for Alzheimer disease (AD), identifying biomarkers for predicting risk for amyloid-related imaging abnormalities (ARIA), hemorrhagic or edema types, is of increased interest. ARIA are thought to be related to disruption of the blood-brain barrier as fibrillary amyloid is cleared from the brain. Molecular and cellular processes related to these events may inform future trials. We investigated proteomics related to abnormal neurovascular imaging phenotypes such as white matter hyperintensities (WMH) in autosomal dominant AD (ADAD), a relatively young population at risk for ARIA.</p> </section> <section> <h3> Methods</h3> <p>Participants from the Dominantly Inherited Alzheimer Network observational study (n<sub>Carriers</sub>=290 and n<sub>Non-Carriers</sub>=183) were assessed for WMH and microhemorrhages using T2-FLAIR and T2*GRE MRI, and for CSF proteomics using the 7k Somalogic<sup>®</sup> platform. A subset (n<sub>Carriers</sub>=92 and n<sub>Non-Carriers</sub>=51) was evaluated for microhemorrhage incidence. WMH volumes were segmented with Triplanar U-Net ensemble network. Microhemorrhage count and incidence were classified as none, mild, moderate, or severe, based on current FDA recommendations. We performed differential abundance analyses to investigate proteins associated with WMH as a function of mutation status, accounting for age, APOE-e4 status, and sex, and significant proteins were further evaluated in pathway analyses and for associations with microhemorrhages.</p> </section> <section> <h3> Results</h3> <p>Eight proteins were differently expressed in carriers with larger WMH volumes (Figure 1). The genes of seven proteins (e.g., neurofilament light-chain (NEFL), neurofilament heavy-chain (NEFH), matrix metalloproteinase 12 (MMP12), fibronectin-1 (FN1), periostin (POSTN)) were overly represented in vascular-related disorders such as subarachnoid hemorrhages, transient ischemic attack, or cerebrovascular diseases (Figure 2). CSF levels of NEFL, NEFH, MMP12, fibronectin1, and periostin differed as a function of CMH severity. Especially, NEFL and MMP12 were higher in carriers with severe CMH compared to those with none or mild CMH (Figure 3A). MMP12 levels were particularly high in participants having severe increase in microhemorrhages (Figure 3B). Carriers with high levels of MMP12 may more likely develop new microhemorrhages.</p> </section> <section> <h3> Conclusions</h3> <p>Our findings confirm the contribution of neurofilament light chain in disease p
{"title":"Elevated CSF levels of matrix metalloproteinase-12 as a potential marker for microhemorrhage risk in autosomal dominant Alzheimer disease","authors":"Nelly Joseph-Mathurin, Katherine Gong, Gengsheng Chen, Parinaz Massoumzadeh, Jeremy F. Strain, Laura Ibanez, Brian A. Gordon, Jorge J. Llibre-Guerra, Jason J. Hassenstab, Richard J. Perrin, Chengjie Xiong, Randall J. Bateman, Eric McDade, Tammie L.S. Benzinger, Carlos Cruchaga, Dominantly Inherited Alzheimer Network","doi":"10.1002/alz70856_107566","DOIUrl":"10.1002/alz70856_107566","url":null,"abstract":"<div>\u0000 \u0000 \u0000 <section>\u0000 \u0000 <h3> Background</h3>\u0000 \u0000 <p>With the advent of disease-modifying treatment for Alzheimer disease (AD), identifying biomarkers for predicting risk for amyloid-related imaging abnormalities (ARIA), hemorrhagic or edema types, is of increased interest. ARIA are thought to be related to disruption of the blood-brain barrier as fibrillary amyloid is cleared from the brain. Molecular and cellular processes related to these events may inform future trials. We investigated proteomics related to abnormal neurovascular imaging phenotypes such as white matter hyperintensities (WMH) in autosomal dominant AD (ADAD), a relatively young population at risk for ARIA.</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Methods</h3>\u0000 \u0000 <p>Participants from the Dominantly Inherited Alzheimer Network observational study (n<sub>Carriers</sub>=290 and n<sub>Non-Carriers</sub>=183) were assessed for WMH and microhemorrhages using T2-FLAIR and T2*GRE MRI, and for CSF proteomics using the 7k Somalogic<sup>®</sup> platform. A subset (n<sub>Carriers</sub>=92 and n<sub>Non-Carriers</sub>=51) was evaluated for microhemorrhage incidence. WMH volumes were segmented with Triplanar U-Net ensemble network. Microhemorrhage count and incidence were classified as none, mild, moderate, or severe, based on current FDA recommendations. We performed differential abundance analyses to investigate proteins associated with WMH as a function of mutation status, accounting for age, APOE-e4 status, and sex, and significant proteins were further evaluated in pathway analyses and for associations with microhemorrhages.</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Results</h3>\u0000 \u0000 <p>Eight proteins were differently expressed in carriers with larger WMH volumes (Figure 1). The genes of seven proteins (e.g., neurofilament light-chain (NEFL), neurofilament heavy-chain (NEFH), matrix metalloproteinase 12 (MMP12), fibronectin-1 (FN1), periostin (POSTN)) were overly represented in vascular-related disorders such as subarachnoid hemorrhages, transient ischemic attack, or cerebrovascular diseases (Figure 2). CSF levels of NEFL, NEFH, MMP12, fibronectin1, and periostin differed as a function of CMH severity. Especially, NEFL and MMP12 were higher in carriers with severe CMH compared to those with none or mild CMH (Figure 3A). MMP12 levels were particularly high in participants having severe increase in microhemorrhages (Figure 3B). Carriers with high levels of MMP12 may more likely develop new microhemorrhages.</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Conclusions</h3>\u0000 \u0000 <p>Our findings confirm the contribution of neurofilament light chain in disease p","PeriodicalId":7471,"journal":{"name":"Alzheimer's & Dementia","volume":"21 S2","pages":""},"PeriodicalIF":11.1,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://alz-journals.onlinelibrary.wiley.com/doi/epdf/10.1002/alz70856_107566","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145948372","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Xueying Lyu, Nidhi S. Mundada, Christopher A Brown, Niyousha Sadeghpour, Emily McGrew, Michael Tran Duong, Long Xie, Mengjin Dong, Yue Li, Ilya M. Nasrallah, Laura E.M. Wisse, Paul A. Yushkevich, Sandhitsu R. Das, David A. Wolk
Background The heterogeneity of Alzheimer's disease (AD) and lack of well‐validated markers of non‐AD factors (e.g. TDP‐43) present a substantial challenge for therapeutics. Our prior work showed discordance between tau (T) and neurodegeneration (N) identified non‐AD factors in AD through multi‐modality imaging. Here we tried a simplified approach using plasma ptau217 and medial temporal lobe (MTL) morphometry, given this region's common association with co‐pathologies, particularly LATE‐NC. Method We included 349 ADNI participants (188 cognitively normal, 161 MCI/dementia) with paired T1‐MRI and plasma ptau217. The MTL was segmented into subregions and further parcellated into 100 bilateral super‐points within regional boundaries. T1‐MRI‐derived thickness and amygdala volume represented N, and plasma p ‐Tau217 represented T. T‐N residuals, calculated through regression across super‐points and amygdala, were used for weighted clustering. Result P‐Tau217 showed strong association with MTL atrophy (Figure 1A). Three distinct data‐driven T‐N groups were identified based on mismatch patterns (Figure 1B), including a canonical group (N∼T), a vulnerable group (N>T) with negative residuals primarily in anterior hippocampal and extrahippocampal areas, and a resilient group (N<T) with positive residuals. After clustering, group comparisons were restricted to the AD continuum (i.e. A+). While groups differed in regional volumes (e.g., amygdala), tau severity did not vary (Table 1), suggesting these patterns were not driven by AD pathology. The vulnerable group, displayed greater anterior MTL atrophy aligning with their T‐N residual patterns, while the resilient group had less atrophy in anterior extrahippocampal area (Figure 2A). Outside the MTL, the vulnerable group showed greater anterior limbic atrophy whereas the resilient group showed less (Figure 2B). The T‐N groups differed in Clinical Dementia Rating (CDR) with the vulnerable group having the worst ratings and the resilient group the best (Table 1). Notably, the vulnerable group demonstrated greater baseline memory impairment. Longitudinally, the vulnerable group also declined more severely across multiple cognitive domains while resilient group remained most stable (Figure 2C). Conclusion T‐N mismatch within MTL using MRI and plasma biomarkers revealed groups with varying vulnerability/resilience, with the vulnerable group displaying patterns of atrophy and cognition suggestive of LATE‐NC. It offers a less invasive, cost‐effective method for stratifying individuals for therapeutic interventions.
{"title":"Medial temporal lobe Tau‐Neurodegeneration mismatch from MRI and plasma biomarkers identifies vulnerable and resilient phenotypes with AD","authors":"Xueying Lyu, Nidhi S. Mundada, Christopher A Brown, Niyousha Sadeghpour, Emily McGrew, Michael Tran Duong, Long Xie, Mengjin Dong, Yue Li, Ilya M. Nasrallah, Laura E.M. Wisse, Paul A. Yushkevich, Sandhitsu R. Das, David A. Wolk","doi":"10.1002/alz70856_107051","DOIUrl":"https://doi.org/10.1002/alz70856_107051","url":null,"abstract":"Background The heterogeneity of Alzheimer's disease (AD) and lack of well‐validated markers of non‐AD factors (e.g. TDP‐43) present a substantial challenge for therapeutics. Our prior work showed discordance between tau (T) and neurodegeneration (N) identified non‐AD factors in AD through multi‐modality imaging. Here we tried a simplified approach using plasma ptau217 and medial temporal lobe (MTL) morphometry, given this region's common association with co‐pathologies, particularly LATE‐NC. Method We included 349 ADNI participants (188 cognitively normal, 161 MCI/dementia) with paired T1‐MRI and plasma ptau217. The MTL was segmented into subregions and further parcellated into 100 bilateral super‐points within regional boundaries. T1‐MRI‐derived thickness and amygdala volume represented N, and plasma <jats:italic>p</jats:italic> ‐Tau217 represented T. T‐N residuals, calculated through regression across super‐points and amygdala, were used for weighted clustering. Result P‐Tau217 showed strong association with MTL atrophy (Figure 1A). Three distinct data‐driven T‐N groups were identified based on mismatch patterns (Figure 1B), including a canonical group (N∼T), a vulnerable group (N>T) with negative residuals primarily in anterior hippocampal and extrahippocampal areas, and a resilient group (N<T) with positive residuals. After clustering, group comparisons were restricted to the AD continuum (i.e. A+). While groups differed in regional volumes (e.g., amygdala), tau severity did not vary (Table 1), suggesting these patterns were not driven by AD pathology. The vulnerable group, displayed greater anterior MTL atrophy aligning with their T‐N residual patterns, while the resilient group had less atrophy in anterior extrahippocampal area (Figure 2A). Outside the MTL, the vulnerable group showed greater anterior limbic atrophy whereas the resilient group showed less (Figure 2B). The T‐N groups differed in Clinical Dementia Rating (CDR) with the vulnerable group having the worst ratings and the resilient group the best (Table 1). Notably, the vulnerable group demonstrated greater baseline memory impairment. Longitudinally, the vulnerable group also declined more severely across multiple cognitive domains while resilient group remained most stable (Figure 2C). Conclusion T‐N <jats:italic>mismatch</jats:italic> within MTL using MRI and plasma biomarkers revealed groups with varying vulnerability/resilience, with the vulnerable group displaying patterns of atrophy and cognition suggestive of LATE‐NC. It offers a less invasive, cost‐effective method for stratifying individuals for therapeutic interventions.","PeriodicalId":7471,"journal":{"name":"Alzheimer's & Dementia","volume":"144 1","pages":""},"PeriodicalIF":14.0,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145947363","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Background Disrupted sleep and 24‐hour rhythms contribute to age‐related inflammation, a key pathway influencing vulnerability to chronic disease and mortality in older adults. Previous studies have explored associations between lifestyle measures and inflammation, but additional research is needed to characterize associations between sleep‐wake rhythms and inflammatory cytokines in the context of healthy brain aging. Method 54 cognitively normal older adults (77.2 ± 6.6 years, 57% female) completed blood draws and 14 days of at‐home actigraphy (wrist‐worn accelerometry). Plasma samples were analyzed using the Alamar NULISASeq Inflammation panel. Actigraphy analyses focused on nocturnal sleep efficiency, sleep duration, 24‐hour relative amplitude, interdaily stability, and intradaily variability. Sixty cytokines and other proteins were selected based on their roles in regulating inflammation. Associations between actigraphy measures and inflammatory markers were tested using linear regression with age, sex, and time between collection (2.0 ± 2.8 years) included as covariates. Multiple comparisons corrections were not used. NULISA p ‐tau217 was added as a covariate in sensitivity analyses. Result Worse sleep and 24‐hour rhythms were generally associated with an elevated inflammatory cytokine profile (Figure). Shorter sleep duration was associated with lower levels of IL‐1 receptor‐like 1 and higher levels of IL‐1β. Day‐to‐day regularity of 24‐hour rhythms (interdaily stability) was positively associated with IL‐5 receptor subunit A, and within‐day fragmentation (intradaily variability) was positively associated with colony stimulating factor 2 receptor beta. Worse sleep efficiency was associated with higher fibroblast growth factor 23. Higher relative amplitude of 24‐hour rhythm was associated with lower IL‐6. When p ‐tau217 was added as a covariate, associations remained significant for IL‐1β, IL‐6, IL‐5RA, FGF23, and CSF2RB, but not for IL‐1RL1, suggesting that these associations were independent of Alzheimer's disease pathology. Conclusion These preliminary findings support the hypothesis that disrupted sleep and 24‐hour rhythms are associated with higher pro‐inflammatory and lower anti‐inflammatory cytokines among healthy older adults. Future work is needed to examine how associations between sleep and inflammation impact cognitive aging trajectories.
{"title":"Sleep and 24‐hour rhythm associations with plasma markers of inflammation in aging","authors":"Skylar E Weiss","doi":"10.1002/alz70856_105983","DOIUrl":"https://doi.org/10.1002/alz70856_105983","url":null,"abstract":"Background Disrupted sleep and 24‐hour rhythms contribute to age‐related inflammation, a key pathway influencing vulnerability to chronic disease and mortality in older adults. Previous studies have explored associations between lifestyle measures and inflammation, but additional research is needed to characterize associations between sleep‐wake rhythms and inflammatory cytokines in the context of healthy brain aging. Method 54 cognitively normal older adults (77.2 ± 6.6 years, 57% female) completed blood draws and 14 days of at‐home actigraphy (wrist‐worn accelerometry). Plasma samples were analyzed using the Alamar NULISASeq Inflammation panel. Actigraphy analyses focused on nocturnal sleep efficiency, sleep duration, 24‐hour relative amplitude, interdaily stability, and intradaily variability. Sixty cytokines and other proteins were selected based on their roles in regulating inflammation. Associations between actigraphy measures and inflammatory markers were tested using linear regression with age, sex, and time between collection (2.0 ± 2.8 years) included as covariates. Multiple comparisons corrections were not used. NULISA <jats:italic>p</jats:italic> ‐tau217 was added as a covariate in sensitivity analyses. Result Worse sleep and 24‐hour rhythms were generally associated with an elevated inflammatory cytokine profile (Figure). Shorter sleep duration was associated with lower levels of IL‐1 receptor‐like 1 and higher levels of IL‐1β. Day‐to‐day regularity of 24‐hour rhythms (interdaily stability) was positively associated with IL‐5 receptor subunit A, and within‐day fragmentation (intradaily variability) was positively associated with colony stimulating factor 2 receptor beta. Worse sleep efficiency was associated with higher fibroblast growth factor 23. Higher relative amplitude of 24‐hour rhythm was associated with lower IL‐6. When <jats:italic>p</jats:italic> ‐tau217 was added as a covariate, associations remained significant for IL‐1β, IL‐6, IL‐5RA, FGF23, and CSF2RB, but not for IL‐1RL1, suggesting that these associations were independent of Alzheimer's disease pathology. Conclusion These preliminary findings support the hypothesis that disrupted sleep and 24‐hour rhythms are associated with higher pro‐inflammatory and lower anti‐inflammatory cytokines among healthy older adults. Future work is needed to examine how associations between sleep and inflammation impact cognitive aging trajectories.","PeriodicalId":7471,"journal":{"name":"Alzheimer's & Dementia","volume":"57 1","pages":""},"PeriodicalIF":14.0,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145947389","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Background Apolipoprotein E (APOE) and ethnicity were proved to have strong effect on Alzheimer's disease. However, study on APOE effect on amyloid‐PET in East Asians was limited. Here, we assess the effects of APOE and race/ethnicity interaction effects on amyloid‐positivity and amyloid‐PET among East Asian and Not Hispanic and Latino White people. Method Linear regression model was used to estimate the APOE and APOE/ethnicity interaction on amyloid‐PET among East Asians ( N = 1529) and Not Hispanic and Latino White (ADNI, N = 1259). Logistic generalized estimating equations were used to estimate the APOE/ethnicity interaction effect on frequency of amyloid‐positivity (using cohort‐specific visual check). For estimation of APOE and APOE/ethnicity effect, the APOE ε3/ε3 was used as reference group. Both models were adjusted for age, sex and years of education. Result APOE ε4 alleles were ascociated with higher risk amyloid deposition compared with ε3/ε3 group (ε2/ε4: β=37.68, p <0.001, ε3/ε4: β=20.96, p <0.001, ε4/ε4: β=36.27, p <0.001) among East Asians. However, ε2 alleles showed no protective effect on amyloid deposition in East Asians. For APOE/ethnicity interaction effect (Figure 2), ε4 alleles in East Asians were associated with less amyloid deposition and amyloid positivity risk than Not Hispanic and Latino White population. In contrast, ε2 alleles were associated with higher risk of amyloid deposition and positivity in East Asians than Not Hispanic and Latino White people. Moreover, APOE ε2/ε4 showed similar effect between East Asians and White people on amyloid deposition and positivity. Conclusion Through recent advances in AD‐related genetic cohorts, this study provided the largest‐to‐date overview of the association of APOE with amyloid‐PET risk in East Asians. APOE ε4 alleles were associated with less amyloid deposition risk and APOE ε2 alleles were associated with higher risk than White people. These novel insights are critical to guide AD clinical trial design and research.
载脂蛋白E （APOE）和种族被证明对阿尔茨海默病有很强的影响。然而，APOE对东亚人淀粉样蛋白- PET影响的研究有限。在这里，我们评估了APOE和种族/民族相互作用对东亚、非西班牙裔和拉丁裔白人淀粉样蛋白阳性和淀粉样蛋白PET的影响。方法采用线性回归模型估计东亚人（N = 1529）和非西班牙裔和拉丁裔白人（ADNI, N = 1259）的APOE和APOE/种族对淀粉样蛋白- PET的相互作用。使用Logistic广义估计方程来估计APOE/种族相互作用对淀粉样蛋白阳性频率的影响（使用队列特异性视觉检查）。为估计APOE和APOE/族裔效应，以APOE ε3/ε3为参照组。两个模型都根据年龄、性别和受教育年限进行了调整。结果与ε3/ε3组（ε2/ε4: β=37.68, p <0.001, ε3/ε4: β=20.96, p <0.001, ε4/ε4: β=36.27, p <0.001）相比，APOE ε4等位基因与东亚人淀粉样蛋白沉积风险较高。而ε2等位基因对东亚人淀粉样蛋白沉积无保护作用。对于APOE/种族相互作用效应（图2），东亚人的ε4等位基因与淀粉样蛋白沉积和淀粉样蛋白阳性风险的相关性低于非西班牙裔和拉丁裔白人。相比之下，东亚人的ε2等位基因与淀粉样蛋白沉积和阳性的风险高于非西班牙裔和拉丁裔白人。此外，APOE ε2/ε4在东亚人和白人之间对淀粉样蛋白沉积和阳性表现出相似的影响。通过对AD相关遗传队列研究的最新进展，本研究提供了迄今为止东亚人APOE与淀粉样蛋白PET风险之间关系的最大规模综述。APOE ε4等位基因与淀粉样蛋白沉积风险相关，而APOE ε2等位基因与淀粉样蛋白沉积风险相关。这些新颖的见解对指导阿尔茨海默病的临床试验设计和研究至关重要。
{"title":"Apolipoprotein E and ethnicity interaction effects on amyloid‐PET status among East Asian and Not Hispanic and Latino White people","authors":"Size Li, Qi Huang, Yihui Guan, Jing Zhang, Fang Xie","doi":"10.1002/alz70856_107790","DOIUrl":"https://doi.org/10.1002/alz70856_107790","url":null,"abstract":"Background Apolipoprotein E (APOE) and ethnicity were proved to have strong effect on Alzheimer's disease. However, study on APOE effect on amyloid‐PET in East Asians was limited. Here, we assess the effects of APOE and race/ethnicity interaction effects on amyloid‐positivity and amyloid‐PET among East Asian and Not Hispanic and Latino White people. Method Linear regression model was used to estimate the APOE and APOE/ethnicity interaction on amyloid‐PET among East Asians ( <jats:italic>N</jats:italic> = 1529) and Not Hispanic and Latino White (ADNI, <jats:italic>N</jats:italic> = 1259). Logistic generalized estimating equations were used to estimate the APOE/ethnicity interaction effect on frequency of amyloid‐positivity (using cohort‐specific visual check). For estimation of APOE and APOE/ethnicity effect, the APOE ε3/ε3 was used as reference group. Both models were adjusted for age, sex and years of education. Result APOE ε4 alleles were ascociated with higher risk amyloid deposition compared with ε3/ε3 group (ε2/ε4: β=37.68, <jats:italic>p</jats:italic> <0.001, ε3/ε4: β=20.96, <jats:italic>p</jats:italic> <0.001, ε4/ε4: β=36.27, <jats:italic>p</jats:italic> <0.001) among East Asians. However, ε2 alleles showed no protective effect on amyloid deposition in East Asians. For APOE/ethnicity interaction effect (Figure 2), ε4 alleles in East Asians were associated with less amyloid deposition and amyloid positivity risk than Not Hispanic and Latino White population. In contrast, ε2 alleles were associated with higher risk of amyloid deposition and positivity in East Asians than Not Hispanic and Latino White people. Moreover, APOE ε2/ε4 showed similar effect between East Asians and White people on amyloid deposition and positivity. Conclusion Through recent advances in AD‐related genetic cohorts, this study provided the largest‐to‐date overview of the association of APOE with amyloid‐PET risk in East Asians. APOE ε4 alleles were associated with less amyloid deposition risk and APOE ε2 alleles were associated with higher risk than White people. These novel insights are critical to guide AD clinical trial design and research.","PeriodicalId":7471,"journal":{"name":"Alzheimer's & Dementia","volume":"27 1","pages":""},"PeriodicalIF":14.0,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145938037","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Melissa Lamar, Maude Wagner, Shengwei Zhang, Sue E. Leurgans, Victoria N Poole, Lisa L. Barnes, David A. A. Bennett, David X. Marquez, Julie A Schneider, Konstantinos Arfanakis
Background Structural connectome‐focused neuroimaging provides an important window into the aging brain; however, few studies address the complexities of mapping structural connectivity in late‐life. Of those that do, most either employ ‘lesion‐filling’ approaches to calculate connectivity metrics, or statistically adjust for lesions. These reasonable applications of connectome metrics to older adults with brain abnormalities and/or focal lesions may nonetheless obscure the impact of such lesions on (dis‐) connectivity and cognitive impairment. Methods We applied alternative solutions to 3T neuroimaging data of 1,047 Rush Alzheimer's Disease Center cohort participants [age(years)≈78±7; 61% non‐Latino White]. We used an atlas‐based definition of the path of white matter connections via a structural connectivity‐based atlas, data‐driven network edge selection, and multi‐modal MRI metrics to reveal subtle distinctions in structural connectome integrity. Global cognition was assessed annually using 19 cognitive measures. Results Data‐driven network edge selection resulted in 308 major edges contributing to the overall structural connectome integrity matrix (SCIM). Separate principal component analyses (PCAs) of MRI‐derived transverse relaxation rates (R 2 ), fractional anisotropy, and quantitative susceptibility mapping revealed modality‐specific sub‐networks. For example, the R 2 ‐SCIM PCA revealed four sub‐networks containing U‐fibers within lobes and connections between lobes (Figure 1): Sub‐Network 1 was characterized by R 2 integrity within edges involving most frontal nodes (i.e., grey matter regions), all parietal nodes, and key subcortical structures including the basal ganglia; Sub‐Network 2 involved most albeit slightly different frontal nodes than Sub‐Network 1, nearly all temporal nodes, and key subcortical structures including limbic structures; Sub‐Network 3 was primarily characterized by R 2 integrity of edges involving select parietal and temporal nodes and all occipital nodes; Sub‐Network 4 involved only the cerebellum, basal ganglia and limbic structures. A linear mixed‐effects regression of global cognition containing PCA‐derived weighted composite scores representing each R 2 ‐SCIM sub‐network and relevant confounders demonstrated associations of higher R 2 in Sub‐Networks 1, 2, and 4 with higher cognition at baseline ( p ‐values≤0.045) and associations of higher R 2 in Sub‐Networks 2 and 3 with slower decline in cognition ( p ‐values≤0.02) (Figure 2). Conclusion Our approach to understanding the aging connectome provides a comprehensive assessment of structural connectome integrity, its underlying sub‐networks, and their associations with cognition.
{"title":"Derivation of a structural connectome integrity matrix based on the complex neural circuitry of the aging brain: A multi‐modal perspective applied to cognitive aging","authors":"Melissa Lamar, Maude Wagner, Shengwei Zhang, Sue E. Leurgans, Victoria N Poole, Lisa L. Barnes, David A. A. Bennett, David X. Marquez, Julie A Schneider, Konstantinos Arfanakis","doi":"10.1002/alz70856_105182","DOIUrl":"https://doi.org/10.1002/alz70856_105182","url":null,"abstract":"Background Structural connectome‐focused neuroimaging provides an important window into the aging brain; however, few studies address the complexities of mapping structural connectivity in late‐life. Of those that do, most either employ ‘lesion‐filling’ approaches to calculate connectivity metrics, or statistically adjust for lesions. These reasonable applications of connectome metrics to older adults with brain abnormalities and/or focal lesions may nonetheless obscure the impact of such lesions on (dis‐) connectivity and cognitive impairment. Methods We applied alternative solutions to 3T neuroimaging data of 1,047 Rush Alzheimer's Disease Center cohort participants [age(years)≈78±7; 61% non‐Latino White]. We used an atlas‐based definition of the path of white matter connections via a structural connectivity‐based atlas, data‐driven network edge selection, and multi‐modal MRI metrics to reveal subtle distinctions in structural connectome integrity. Global cognition was assessed annually using 19 cognitive measures. Results Data‐driven network edge selection resulted in 308 major edges contributing to the overall structural connectome integrity matrix (SCIM). Separate principal component analyses (PCAs) of MRI‐derived transverse relaxation rates (R <jats:sub>2</jats:sub> ), fractional anisotropy, and quantitative susceptibility mapping revealed modality‐specific sub‐networks. For example, the R <jats:sub>2</jats:sub> ‐SCIM PCA revealed four sub‐networks containing U‐fibers within lobes and connections between lobes (Figure 1): Sub‐Network 1 was characterized by R <jats:sub>2</jats:sub> integrity within edges involving most frontal nodes (i.e., grey matter regions), all parietal nodes, and key subcortical structures including the basal ganglia; Sub‐Network 2 involved most albeit slightly different frontal nodes than Sub‐Network 1, nearly all temporal nodes, and key subcortical structures including limbic structures; Sub‐Network 3 was primarily characterized by R <jats:sub>2</jats:sub> integrity of edges involving select parietal and temporal nodes and all occipital nodes; Sub‐Network 4 involved only the cerebellum, basal ganglia and limbic structures. A linear mixed‐effects regression of global cognition containing PCA‐derived weighted composite scores representing each R <jats:sub>2</jats:sub> ‐SCIM sub‐network and relevant confounders demonstrated associations of higher R <jats:sub>2</jats:sub> in Sub‐Networks 1, 2, and 4 with higher cognition at baseline ( <jats:italic>p</jats:italic> ‐values≤0.045) and associations of higher R <jats:sub>2</jats:sub> in Sub‐Networks 2 and 3 with slower decline in cognition ( <jats:italic>p</jats:italic> ‐values≤0.02) (Figure 2). Conclusion Our approach to understanding the aging connectome provides a comprehensive assessment of structural connectome integrity, its underlying sub‐networks, and their associations with cognition.","PeriodicalId":7471,"journal":{"name":"Alzheimer's & Dementia","volume":"55 1","pages":""},"PeriodicalIF":14.0,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145938078","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：