Epigenetic and genetic risk of Alzheimer disease from autopsied brains in two ethnic groups.

Genetic variants and epigenetic features both contribute to the risk of Alzheimer's disease (AD). We studied the AD association of CpG-related single nucleotide polymorphisms (CGS), which act as a hub of both the genetic and epigenetic effects, in Caribbean Hispanics (CH) and generalized the findings to Non-Hispanic Whites (NHW). First, we conducted a genome-wide, sliding-window-based association with AD, in 7,155 CH and 1,283 NHW participants. Next, using data from the dorsolateral prefrontal cortex in 179 CH brains, we tested the cis- and trans-effects of AD-associated CGS on brain DNA methylation to mRNA expression. For the genes with significant cis- and trans-effects, we investigated their enriched pathways. We identified six genetic loci in CH with CGS dosage associated with AD at genome-wide significance levels: ADAM20 (Score = 55.19, P = 4.06 × 10-8), the intergenic region between VRTN and SYNDIG1L (Score = - 37.67, P = 2.25 × 10-9), SPG7 (16q24.3) (Score = 40.51, P = 2.23 × 10-8), PVRL2 (Score = 125.86, P = 1.64 × 10-9), TOMM40 (Score = - 18.58, P = 4.61 × 10-8), and APOE (Score = 75.12, P = 7.26 × 10-26). CGSes in PVRL2 and APOE were also significant in NHW. Except for ADAM20, CGSes in the other five loci were associated with CH brain methylation levels (mQTLs) and CGSes in SPG7, PVRL2, and APOE were also mQTLs in NHW. Except for SYNDIG1L (P = 0.08), brain methylation levels in the other five loci affected downstream mRNA expression in CH (P < 0.05), and methylation at VRTN and TOMM40 were also associated with mRNA expression in NHW. Gene expression in these six loci were also regulated by CpG sites in genes that were enriched in the neuron projection and glutamatergic synapse pathways (FDR < 0.05). DNA methylation at all six loci and mRNA expression of SYNDIG1 and TOMM40 were significantly associated with Braak Stage in CH. In summary, we identified six CpG-related genetic loci associated with AD in CH, harboring both genetic and epigenetic risks. However, their downstream effects on mRNA expression maybe ethnic specific and different from NHW.

LATE-NC staging in routine neuropathologic diagnosis: an update.

An international consensus report in 2019 recommended a classification system for limbic-predominant age-related TDP-43 encephalopathy neuropathologic changes (LATE-NC). The suggested neuropathologic staging system and nomenclature have proven useful for autopsy practice and dementia research. However, some issues remain unresolved, such as cases with unusual features that do not fit with current diagnostic categories. The goal of this report is to update the neuropathologic criteria for the diagnosis and staging of LATE-NC, based primarily on published data. We provide practical suggestions about how to integrate available genetic information and comorbid pathologies [e.g., Alzheimer's disease neuropathologic changes (ADNC) and Lewy body disease]. We also describe recent research findings that have enabled more precise guidance on how to differentiate LATE-NC from other subtypes of TDP-43 pathology [e.g., frontotemporal lobar degeneration (FTLD) and amyotrophic lateral sclerosis (ALS)], and how to render diagnoses in unusual situations in which TDP-43 pathology does not follow the staging scheme proposed in 2019. Specific recommendations are also made on when not to apply this diagnostic term based on current knowledge. Neuroanatomical regions of interest in LATE-NC are described in detail and the implications for TDP-43 immunohistochemical results are specified more precisely. We also highlight questions that remain unresolved and areas needing additional study. In summary, the current work lays out a number of recommendations to improve the precision of LATE-NC staging based on published reports and diagnostic experience.

Frequency of LATE neuropathologic change across the spectrum of Alzheimer's disease neuropathology: combined data from 13 community-based or population-based autopsy cohorts.

Limbic-predominant age-related TDP-43 encephalopathy neuropathologic change (LATE-NC) and Alzheimer's disease neuropathologic change (ADNC) are each associated with substantial cognitive impairment in aging populations. However, the prevalence of LATE-NC across the full range of ADNC remains uncertain. To address this knowledge gap, neuropathologic, genetic, and clinical data were compiled from 13 high-quality community- and population-based longitudinal studies. Participants were recruited from United States (8 cohorts, including one focusing on Japanese-American men), United Kingdom (2 cohorts), Brazil, Austria, and Finland. The total number of participants included was 6196, and the average age of death was 88.1 years. Not all data were available on each individual and there were differences between the cohorts in study designs and the amount of missing data. Among those with known cognitive status before death (n = 5665), 43.0% were cognitively normal, 14.9% had MCI, and 42.4% had dementia-broadly consistent with epidemiologic data in this age group. Approximately 99% of participants (n = 6125) had available CERAD neuritic amyloid plaque score data. In this subsample, 39.4% had autopsy-confirmed LATE-NC of any stage. Among brains with "frequent" neuritic amyloid plaques, 54.9% had comorbid LATE-NC, whereas in brains with no detected neuritic amyloid plaques, 27.0% had LATE-NC. Data on LATE-NC stages were available for 3803 participants, of which 25% had LATE-NC stage > 1 (associated with cognitive impairment). In the subset of individuals with Thal Aß phase = 0 (lacking detectable Aß plaques), the brains with LATE-NC had relatively more severe primary age-related tauopathy (PART). A total of 3267 participants had available clinical data relevant to frontotemporal dementia (FTD), and none were given the clinical diagnosis of definite FTD nor the pathological diagnosis of frontotemporal lobar degeneration with TDP-43 inclusions (FTLD-TDP). In the 10 cohorts with detailed neurocognitive assessments proximal to death, cognition tended to be worse with LATE-NC across the full spectrum of ADNC severity. This study provided a credible estimate of the current prevalence of LATE-NC in advanced age. LATE-NC was seen in almost 40% of participants and often, but not always, coexisted with Alzheimer's disease neuropathology.

Neuropathological Findings in Parkinson's Disease With Mild Cognitive Impairment.

OBJECTIVE: There are few neuropathological studies on Parkinson's disease with mild cognitive impairment (PD-MCI). Those published reveal coexisting Lewy body and Alzheimer's disease pathology. Our objective is to determine the pathology that underlies PD-MCI. METHODS: We used data from the Arizona Study of Aging and Neurodegenerative Disorders, a longitudinal clinicopathological study. Of 736 autopsied subjects with standardized movement and cognitive assessments, 25 had PD-MCI. Neuropathological findings, including Lewy body and Alzheimer's disease pathology, were compared in PD subjects with amnestic MCI (A-MCI) and nonamnestic MCI (NA-MCI). RESULTS: Significant pathological heterogeneity within PD-MCI was found. This included varying Lewy body stages, Alzheimer's disease pathology, and cerebral amyloid angiopathy. There was a significant increase in the severity of Lewy body pathology (meeting The Unified Staging System for Lewy Body disorders neocortical stage) in nonamnestic MCI (7/1, 63%) when compared with amnestic MCI (3/14, 21%, P = 0.032). CONCLUSION: Although a small study, distinct pathological changes may contribute to PD-MCI phenotype. © 2020 International Parkinson and Movement Disorder Society.

Lewy body pathology in Alzheimer's disease: A clinicopathological prospective study.

OBJECTIVE: Identify clinical features predictive of Lewy body pathology in Alzheimer's disease (AD) patients in an ongoing longitudinal clinicopathologic study. MATERIAL AND METHODS: We queried the Arizona Study of Aging and Neurodegenerative Disorders (AZSAND) database for dementia cases with AD pathology (1997-2015). Subjects received longitudinal comprehensive clinical evaluations including motor/neuropsychological assessment and Apo-E4 genotyping. All cases were autopsied and had standard neuropathological assessments for AD and Lewy-type synucleinopathy (LTS). Subjects were categorized based on standardized pathological criteria with AD cases that had LTS but did not meet DLB pathologic criteria being categorized as ADLB. We performed pairwise comparison between the different diagnoses and multivariable modelling to identify clinical symptoms that predict the pathological diagnosis. RESULTS: We identified 32 DLB/AD, 54 ADLB, 70 AD only and 41 PDD/AD cases. AD subjects with LTS pathology had higher UPDRS II and III total scores as well as generally higher individual scores compared to AD alone. While depression scales and Trail-making Test A correlated significantly with LTS, other neuropsychological variables were not significantly different. Apo E4 occurrence was similar in all groups (40%-49%). CONCLUSIONS: Our study suggests that the presence (or absence) of LTS influences motor and non-motor clinical findings in AD patients. These findings may lead to biomarkers that allow for more targeted treatment of AD.

Guinea Pig Prion Protein Supports Rapid Propagation of Bovine Spongiform Encephalopathy and Variant Creutzfeldt-Jakob Disease Prions.

The biochemical and neuropathological properties of bovine spongiform encephalopathy (BSE) and variant Creutzfeldt-Jakob disease (vCJD) prions are faithfully maintained upon transmission to guinea pigs. However, primary and secondary transmissions of BSE and vCJD in guinea pigs result in long incubation periods of ∼450 and ∼350 days, respectively. To determine if the incubation periods of BSE and vCJD prions could be shortened, we generated transgenic (Tg) mice expressing guinea pig prion protein (GPPrP). Inoculation of Tg(GPPrP) mice with BSE and vCJD prions resulted in mean incubation periods of 210 and 199 days, respectively, which shortened to 137 and 122 days upon serial transmission. In contrast, three different isolates of sporadic CJD prions failed to transmit disease to Tg(GPPrP) mice. Many of the strain-specified biochemical and neuropathological properties of BSE and vCJD prions, including the presence of type 2 protease-resistant PrPSc, were preserved upon propagation in Tg(GPPrP) mice. Structural modeling revealed that two residues near the N-terminal region of α-helix 1 in GPPrP might mediate its susceptibility to BSE and vCJD prions. Our results demonstrate that expression of GPPrP in Tg mice supports the rapid propagation of BSE and vCJD prions and suggest that Tg(GPPrP) mice may serve as a useful paradigm for bioassaying these prion isolates. IMPORTANCE: Variant Creutzfeldt-Jakob disease (vCJD) and bovine spongiform encephalopathy (BSE) prions are two of the prion strains most relevant to human health. However, propagating these strains in mice expressing human or bovine prion protein has been difficult because of prolonged incubation periods or inefficient transmission. Here, we show that transgenic mice expressing guinea pig prion protein are fully susceptible to vCJD and BSE prions but not to sporadic CJD prions. Our results suggest that the guinea pig prion protein is a better, more rapid substrate than either bovine or human prion protein for propagating BSE and vCJD prions.

Optimization of Aryl Amides that Extend Survival in Prion-Infected Mice.

Developing therapeutics for neurodegenerative diseases (NDs) prevalent in the aging population remains a daunting challenge. With the growing understanding that many NDs progress by conformational self-templating of specific proteins, the prototypical prion diseases offer a platform for ND drug discovery. We evaluated high-throughput screening hits with the aryl amide scaffold and explored the structure-activity relationships around three series differing in their N-aryl core: benzoxazole, benzothiazole, and cyano. Potent anti-prion compounds were advanced to pharmacokinetic studies, and the resulting brain-penetrant leads from each series, together with a related N-aryl piperazine lead, were escalated to long-term dosing and efficacy studies. Compounds from each of the four series doubled the survival of mice infected with a mouse-passaged prion strain. Treatment with aryl amides altered prion strain properties, as evidenced by the distinct patterns of neuropathological deposition of prion protein and associated astrocytic gliosis in the brain; however, none of the aryl amide compounds resulted in drug-resistant prion strains, in contrast to previous studies on compounds with the 2-aminothiazole (2-AMT) scaffold. As seen with 2-AMTs and other effective anti-prion compounds reported to date, the novel aryl amides reported here were ineffective in prolonging the survival of transgenic mice infected with human prions. Most encouraging is our discovery that aryl amides show that the development of drug resistance is not an inevitable consequence of efficacious anti-prion therapeutics.

Aging-related tau astrogliopathy (ARTAG): harmonized evaluation strategy.

Pathological accumulation of abnormally phosphorylated tau protein in astrocytes is a frequent, but poorly characterized feature of the aging brain. Its etiology is uncertain, but its presence is sufficiently ubiquitous to merit further characterization and classification, which may stimulate clinicopathological studies and research into its pathobiology. This paper aims to harmonize evaluation and nomenclature of aging-related tau astrogliopathy (ARTAG), a term that refers to a morphological spectrum of astroglial pathology detected by tau immunohistochemistry, especially with phosphorylation-dependent and 4R isoform-specific antibodies. ARTAG occurs mainly, but not exclusively, in individuals over 60 years of age. Tau-immunoreactive astrocytes in ARTAG include thorn-shaped astrocytes at the glia limitans and in white matter, as well as solitary or clustered astrocytes with perinuclear cytoplasmic tau immunoreactivity that extends into the astroglial processes as fine fibrillar or granular immunopositivity, typically in gray matter. Various forms of ARTAG may coexist in the same brain and might reflect different pathogenic processes. Based on morphology and anatomical distribution, ARTAG can be distinguished from primary tauopathies, but may be concurrent with primary tauopathies or other disorders. We recommend four steps for evaluation of ARTAG: (1) identification of five types based on the location of either morphologies of tau astrogliopathy: subpial, subependymal, perivascular, white matter, gray matter; (2) documentation of the regional involvement: medial temporal lobe, lobar (frontal, parietal, occipital, lateral temporal), subcortical, brainstem; (3) documentation of the severity of tau astrogliopathy; and (4) description of subregional involvement. Some types of ARTAG may underlie neurological symptoms; however, the clinical significance of ARTAG is currently uncertain and awaits further studies. The goal of this proposal is to raise awareness of astroglial tau pathology in the aged brain, facilitating communication among neuropathologists and researchers, and informing interpretation of clinical biomarkers and imaging studies that focus on tau-related indicators.

Peripheral Synucleinopathy in Early Parkinson's Disease: Submandibular Gland Needle Biopsy Findings.

INTRODUCTION: Finding a peripheral tissue biopsy site to diagnose early PD would be of value for clinical care, biomarker validation, and as research enrollment criteria. Whereas autopsy and advanced PD studies suggest that the submandibular gland is an important biopsy site, there are no studies in early PD. The aim of this study was to determine whether needle biopsy of the submandibular gland reveals Lewy type alpha-synucleinopathy in early PD. METHODS: Twenty-five early PD (duration < 5 years) and 10 controls underwent transcutaneous needle core biopsies of the submandibular gland. Tissue was stained for phosphorylated alpha-synuclein, reviewed blind to clinical diagnosis, and only nerve element staining was considered positive. RESULTS: Mean (standard deviation) age was 69.5 (8.3) for the PD group, 64.8 (8.0) years for controls, and disease duration 2.6 (1.1) years. Six PD and 1 control subject had inadequate glandular tissue. Positive staining was found in 14 of 19 (74%) PD and 2 of 9 (22%) control subjects. PD-positive and -negative cases did not differ clinically. Adverse events (mainly swelling and bruising) were common (77% of cases), but were minor and transient. CONCLUSIONS: Submandibular gland needle biopsies identified phosphorylated alpha-synuclein staining in 74% of early PD subjects. False positives may be true false positives or may represent prodromal PD. If confirmed in larger studies with eventual autopsy confirmation, the potential value of submandibular gland biopsies for early PD may be to aid in clinical trial inclusion/exclusion and eventually serve as a gold standard for biomarker studies short of autopsy confirmation.

A Cross-Sectional Analysis of Late-Life Cardiovascular Factors and Their Relation to Clinically Defined Neurodegenerative Diseases.

Studies have demonstrated associations between cardiovascular factors and Alzheimer disease (AD) with minimal focus on other neurodegenerative diseases. Utilizing cross-sectional data from 17,532 individuals in the National Alzheimer's Coordinating Center, Uniform Data Set, we compared the presence of cardiovascular factors [body mass index (BMI), atrial fibrillation, hypertension, hyperlipidemia, and diabetes] in individuals carrying a diagnosis of Probable AD (ProbAD), Possible AD, vascular dementia, dementia with Lewy bodies (DLB), frontotemporal dementia, Parkinson disease, progressive supranuclear palsy, or corticobasal degeneration, with that of normals. Generalized linear mixed models were fitted with age at visit, gender, and cardiovascular factors as fixed effects and Alzheimer's Disease Centers as random effects. In late life, only BMI of ProbAD and DLB patients was statistically significantly lower than that in normals (P-values <0.001). When accounting for colinearity within cardiovascular factors, a low BMI was a comorbidity of certain dementia etiologies as compared with normals. These data support a concept of disease-specific associations with certain cardiovascular factors.

Description and cohort characterization of the Longevity Study: learning from our elders.

The purpose of this article is to describe the Longevity Study: Learning From Our Elders, a research program on healthy aging that began in 2007 at the Center for Healthy Aging at Banner Sun Health Research Institute. As of June 2015, 1139 participants (age range of 50-110 years) completed baseline assessments with the majority living in the Sun Cities retirement communities northwest of Phoenix, Arizona but expanding throughout the state. The registry includes over 830 currently active participants with 450 aged 80 years and older, 188 aged 90 and older, and 27 centenarians. Data from in-person interviews at the Center for Healthy Aging in Sun City or in the participants' residences which includes sociodemographic, medical, cognitive, physical and psychosocial variables have been collected since the study's inception. This paper outlines some of the key demographic and clinical characteristics of the Longevity Study, its progress, and future directions. It also reflects on how exceptional aging individuals function psychosocially, cognitively and physically, particularly among individuals aged 85 and older.

REM sleep behavior disorder and neuropathology in Parkinson's disease.

INTRODUCTION: Rapid eye movement (REM) sleep behavior disorder (RBD) in Parkinson's disease (PD) is associated with differences in clinical phenotype, including dementia, autonomic loss, and gait dysfunction. The pathological basis for this remains unclear. METHODS: Parkinson's disease subjects in a longitudinal clinicopathologic study were screened for probable RBD with the Mayo Sleep Questionnaire. After death, semiquantitative analyses were conducted for synuclein, amyloid, neurofibrillary tangles, and cerebrovascular lesions. RESULTS: Forty cases had probable RBD (PD+RBD), and 41 did not (PD-RBD). Despite similar age at death (∼80 y) and disease duration (∼14.5 y), PD+RBD had increased synuclein deposition in all regions examined, with nine of 10 regions significantly different. The Lewy body 10-region total score (scale = 0-40) was 29.5 in PD+RBD versus 24.5 in PD-RBD (Cohen-d effect size = 0.79, P = 0.002). Cerebrovascular lesion burden was slightly higher in PD-RBD. CONCLUSIONS: Although overlap occurs between groups, PD patients with probable RBD may have greater density and range of synuclein pathology on autopsy.

Neuropathological comparisons of amnestic and nonamnestic mild cognitive impairment.

BACKGROUND: Although there are studies investigating the pathologic origins of mild cognitive impairment (MCI), they have revolved around comparisons to normal elderly individuals or those with Alzheimer's disease (AD) or other dementias. There are few studies directly comparing the comprehensive neuropathology of amnestic (aMCI) and nonamnestic (naMCI) MCI. METHODS: The database of the Brain and Body Donation Program ( www.brainandbodydonationprogram.org ), a longitudinal clinicopathological study of normal aging and neurodegenerative disorders, was queried for subjects who were carrying a diagnosis of aMCI or naMCI at the time of autopsy. Neuropathological lesions, including neuritic plaques, neurofibrillary tangles (NFTs), Lewy bodies (LBs), infarcts, cerebral white matter rarefaction (CWMR), cerebral amyloid angiopathy (CAA), and concurrent major clinicopathological diagnoses, including Parkinson's disease (PD) were analyzed. RESULTS: Thirty four subjects with aMCI and 15 naMCI met study criteria. Subjects with aMCI were older at death (88 vs. 83 years of age, p = 0.03). Individuals with naMCI had higher densities of LBs within the temporal lobe (p = 0.04) while subjects with aMCI had a propensity for increased NFTs in parietal and temporal lobes (p values = 0.07). After adjusting for age at death, the only significant difference was greater densities of temporal lobe NFTs within the aMCI group. Other regional pathology scores for plaques, NFTs, and LBs were similar between groups. Subjects met clinico-pathological criteria for co-existent PD in 24 % aMCI and 47 % naMCI while neuropathological criteria for AD were met in equal percentages of aMCI and of naMCI cases (53 %); these proportional differences were not significant (p values > 0.35). Furthermore, regardless of amnestic status, there was a greater presence of CAA (71 % of MCI with executive dysfunction vs. 39 % without p = 0.03) and a greater presence of CWMR (81 % of MCI with executive dysfunction and 54 % without p = 0.046) in MCI cases with executive dysfunction. CONCLUSIONS: No single pathologic entity strongly dichotomized MCI groups, perhaps due to the pathologic heterogeneity found within both entities. However, these data suggest the possibility for naMCI to have a propensity for increased LBs and aMCI for increased NFTs in select anatomic regions.

Arizona Study of Aging and Neurodegenerative Disorders and Brain and Body Donation Program.

The Brain and Body Donation Program (BBDP) at Banner Sun Health Research Institute (http://www.brainandbodydonationprogram.org) started in 1987 with brain-only donations and currently has banked more than 1600 brains. More than 430 whole-body donations have been received since this service was commenced in 2005. The collective academic output of the BBDP is now described as the Arizona Study of Aging and Neurodegenerative Disorders (AZSAND). Most BBDP subjects are enrolled as cognitively normal volunteers residing in the retirement communities of metropolitan Phoenix, Arizona. Specific recruitment efforts are also directed at subjects with Alzheimer's disease, Parkinson's disease and cancer. The median age at death is 82. Subjects receive standardized general medical, neurological, neuropsychological and movement disorders assessments during life and more than 90% receive full pathological examinations by medically licensed pathologists after death. The Program has been funded through a combination of internal, federal and state of Arizona grants as well as user fees and pharmaceutical industry collaborations. Subsets of the Program are utilized by the US National Institute on Aging Arizona Alzheimer's Disease Core Center and the US National Institute of Neurological Disorders and Stroke National Brain and Tissue Resource for Parkinson's Disease and Related Disorders. Substantial funding has also been received from the Michael J. Fox Foundation for Parkinson's Research. The Program has made rapid autopsy a priority, with a 3.0-hour median post-mortem interval for the entire collection. The median RNA Integrity Number (RIN) for frozen brain and body tissue is 8.9 and 7.4, respectively. More than 2500 tissue requests have been served and currently about 200 are served annually. These requests have been made by more than 400 investigators located in 32 US states and 15 countries. Tissue from the BBDP has contributed to more than 350 publications and more than 200 grant-funded projects.

Machine learning quantification of Amyloid-ß deposits in the temporal lobe of 131 brain bank cases.

Accurate and scalable quantification of amyloid-ß (Aß) pathology is crucial for deeper disease phenotyping and furthering research in Alzheimer Disease (AD). This multidisciplinary study addresses the current limitations on neuropathology by leveraging a machine learning (ML) pipeline to perform a granular quantification of Aß deposits and assess their distribution in the temporal lobe. Utilizing 131 whole-slide-images from consecutive autopsied cases at the University of California Davis Alzheimer Disease Research Center, our objectives were threefold: (1) Validate an automatic workflow for Aß deposit quantification in white matter (WM) and gray matter (GM); (2) define the distributions of different Aß deposit types in GM and WM, and (3) investigate correlates of Aß deposits with dementia status and the presence of mixed pathology. Our methodology highlights the robustness and efficacy of the ML pipeline, demonstrating proficiency akin to experts' evaluations. We provide comprehensive insights into the quantification and distribution of Aß deposits in the temporal GM and WM revealing a progressive increase in tandem with the severity of established diagnostic criteria (NIA-AA). We also present correlations of Aß load with clinical diagnosis as well as presence/absence of mixed pathology. This study introduces a reproducible workflow, showcasing the practical use of ML approaches in the field of neuropathology, and use of the output data for correlative analyses. Acknowledging limitations, such as potential biases in the ML model and current ML classifications, we propose avenues for future research to refine and expand the methodology. We hope to contribute to the broader landscape of neuropathology advancements, ML applications, and precision medicine, paving the way for deep phenotyping of AD brain cases and establishing a foundation for further advancements in neuropathological research.

Learning precise segmentation of neurofibrillary tangles from rapid manual point annotations.

Accumulation of abnormal tau protein into neurofibrillary tangles (NFTs) is a pathologic hallmark of Alzheimer disease (AD). Accurate detection of NFTs in tissue samples can reveal relationships with clinical, demographic, and genetic features through deep phenotyping. However, expert manual analysis is time-consuming, subject to observer variability, and cannot handle the data amounts generated by modern imaging. We present a scalable, open-source, deep-learning approach to quantify NFT burden in digital whole slide images (WSIs) of post-mortem human brain tissue. To achieve this, we developed a method to generate detailed NFT boundaries directly from single-point-per-NFT annotations. We then trained a semantic segmentation model on 45 annotated 2400µm by 1200µm regions of interest (ROIs) selected from 15 unique temporal cortex WSIs of AD cases from three institutions (University of California (UC)-Davis, UC-San Diego, and Columbia University). Segmenting NFTs at the single-pixel level, the model achieved an area under the receiver operating characteristic of 0.832 and an F1 of 0.527 (196-fold over random) on a held-out test set of 664 NFTs from 20 ROIs (7 WSIs). We compared this to deep object detection, which achieved comparable but coarser-grained performance that was 60% faster. The segmentation and object detection models correlated well with expert semi-quantitative scores at the whole-slide level (Spearman's rho ρ=0.654 (p=6.50e-5) and ρ=0.513 (p=3.18e-3), respectively). We openly release this multi-institution deep-learning pipeline to provide detailed NFT spatial distribution and morphology analysis capability at a scale otherwise infeasible by manual assessment.