
==== Front
Acta Neuropathol
Acta Neuropathol
Acta Neuropathologica
0001-6322
1432-0533
Springer Berlin Heidelberg Berlin/Heidelberg

39227502
2789
10.1007/s00401-024-02789-9
Original Paper
Association of quantitative histopathology measurements with antemortem medial temporal lobe cortical thickness in the Alzheimer’s disease continuum
http://orcid.org/0009-0005-2531-9346
Denning Amanda E. amanda.denning@pennmedicine.upenn.edu

1
Ittyerah Ranjit 1
Levorse Lisa M. 1
Sadeghpour Niyousha 1
Athalye Chinmayee 2
Chung Eunice 1
Ravikumar Sadhana 1
Dong Mengjin 1
Duong Michael Tran 2
Li Yue 1
Ilesanmi Ademola 3
Sreepada Lasya P. 23
Sabatini Philip 3
Lowe MaKayla 3
Bahena Alejandra 3
Zablah Jamila 3
Spencer Barbara E. 3
Watanabe Ryohei 45
Kim Boram 45
Sørensen Maja Højvang 45
Khandelwal Pulkit 2
Brown Christopher 3
Hrybouski Stanislau 1
Xie Sharon X. 6
de Flores Robin 7
Robinson John L. 45
Schuck Theresa 45
Ohm Daniel T. 3
Arezoumandan Sanaz 3
Porta Sílvia 45
Detre John A. 3
Insausti Ricardo 8
Wisse Laura E. M. 9
Das Sandhitsu R. 3
Irwin David J. 3
Lee Edward B. 45
Wolk David A. 3
Yushkevich Paul A. 1
1 https://ror.org/00b30xv10 grid.25879.31 0000 0004 1936 8972 Department of Radiology, University of Pennsylvania, Philadelphia, PA USA
2 https://ror.org/00b30xv10 grid.25879.31 0000 0004 1936 8972 Department of Bioengineering, University of Pennsylvania, Philadelphia, PA USA
3 https://ror.org/00b30xv10 grid.25879.31 0000 0004 1936 8972 Department of Neurology, University of Pennsylvania, Philadelphia, PA USA
4 https://ror.org/00b30xv10 grid.25879.31 0000 0004 1936 8972 Department of Pathology and Laboratory Medicine, University of Pennsylvania, Philadelphia, PA USA
5 grid.25879.31 0000 0004 1936 8972 Center for Neurodegenerative Disease Research, Institute On Aging, University of Pennsylvania, Philadelphia, PA USA
6 https://ror.org/00b30xv10 grid.25879.31 0000 0004 1936 8972 Department of Biostatistics, Epidemiology and Informatics, University of Pennsylvania, Philadelphia, PA USA
7 grid.417831.8 0000 0004 0640 679X UMR-S U1237, PhIND “Physiopathology and Imaging of Neurological Disorders”, Institut Blood and Brain @ Caen-Normandie, INSERM, Caen-Normandie University, GIP Cyceron, Caen, France
8 https://ror.org/05r78ng12 grid.8048.4 0000 0001 2194 2329 Human Neuroanatomy Lab, University of Castilla La Mancha, Albacete, Spain
9 https://ror.org/012a77v79 grid.4514.4 0000 0001 0930 2361 Department of Clinical Sciences Lund, Lund University, Lund, Sweden
3 9 2024
3 9 2024
2024
148 1 378 5 2024
7 8 2024
15 8 2024
© The Author(s) 2024
2024
https://creativecommons.org/licenses/by/4.0/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
The medial temporal lobe (MTL) is a hotspot for neuropathology, and measurements of MTL atrophy are often used as a biomarker for cognitive decline associated with neurodegenerative disease. Due to the aggregation of multiple proteinopathies in this region, the specific relationship of MTL atrophy to distinct neuropathologies is not well understood. Here, we develop two quantitative algorithms using deep learning to measure phosphorylated tau (p-tau) and TDP-43 (pTDP-43) pathology, which are both known to accumulate in the MTL and are associated with MTL neurodegeneration. We focus on these pathologies in the context of Alzheimer’s disease (AD) and limbic predominant age-related TDP-43 encephalopathy (LATE) and apply our deep learning algorithms to distinct histology sections, on which MTL subregions were digitally annotated. We demonstrate that both quantitative pathology measures show high agreement with expert visual ratings of pathology and discriminate well between pathology stages. In 140 cases with antemortem MR imaging, we compare the association of semi-quantitative and quantitative postmortem measures of these pathologies in the hippocampus with in vivo structural measures of the MTL and its subregions. We find widespread associations of p-tau pathology with MTL subregional structural measures, whereas pTDP-43 pathology had more limited associations with the hippocampus and entorhinal cortex. Quantitative measurements of p-tau pathology resulted in a significantly better model of antemortem structural measures than semi-quantitative ratings and showed strong associations with cortical thickness and volume. By providing a more granular measure of pathology, the quantitative p-tau measures also showed a significant negative association with structure in a severe AD subgroup where semi-quantitative ratings displayed a ceiling effect. Our findings demonstrate the advantages of using quantitative neuropathology to understand the relationship of pathology to structure, particularly for p-tau, and motivate the use of quantitative pathology measurements in future studies.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00401-024-02789-9.

Keywords

Alzheimer’s disease
Quantitative neuropathology
Tau
TDP-43
http://dx.doi.org/10.13039/100000049 National Institute on Aging R01 AG069474 RF1 AG056014 P30 AG072979 P01 AG066597 U19 AG062418 R01 AG075276 R01 AG070592 T32 AG076411 Porta Sílvia Detre John A. Wisse Laura E. M. Lee Edward B. Yushkevich Paul A. http://dx.doi.org/10.13039/501100003252 Lunds Universitet http://dx.doi.org/10.13039/100000065 National Institute of Neurological Disorders and Stroke R25-NS065745 R01-NS109260 Detre John A. Irwin David J. http://dx.doi.org/10.13039/100000957 Alzheimer's Association AACSF-23-1152241 AARF-21- 48972 Brown Christopher Hrybouski Stanislau issue-copyright-statement© Springer-Verlag GmbH Germany, part of Springer Nature 2024
==== Body
pmcIntroduction

The medial temporal lobe (MTL) is an epicenter of pathology and atrophy in multiple neurodegenerative diseases. Measurements of thickness and volume of the MTL and its subregions derived from in vivo magnetic resonance imaging (MRI) are associated with cognitive decline and can serve as effective biomarkers of disease progression [40, 66, 87, 97, 101]. Advancements in MRI and image processing techniques have enabled in vivo imaging and quantitative analysis of the anatomically and functionally distinct subfields/subregions that make up the hippocampus and MTL [36, 74, 101, 104]. Volume and thickness measurements of these subregions have been used in aging and neurodegenerative disease studies examining longitudinal subregional atrophy patterns, functional connectivity, and positron emission tomography (PET) imaging of amyloid and tau pathology [3, 13, 34, 76, 100].

However, multiple neurodegenerative pathologies contribute to neurodegeneration in the MTL, often concurrently, including those involved in Alzheimer’s disease (AD), Lewy Body Disease (LBD), hippocampal sclerosis (HS), and limbic predominant age-related TDP-43 encephalopathy (LATE) [27, 43, 55, 77–79, 83, 96, 102]. Therefore, while measurements of MTL volume and cortical thickness can reflect neurodegeneration associated with neurodegenerative disease in general, the specificity of these patterns as downstream effects of distinct neuropathologies is not well understood. With the recent availability of disease-modifying treatments for AD [23, 82], there is increased need to measure the likelihood of non-AD pathology in individual patients, as this may impact decisions to start treatment as well as the degree of treatment-response. Targeted in vivo biomarkers for non-AD neuropathologies are currently lacking. Fluid biomarkers for alpha-synuclein [30] and TDP-43 [12] are promising, but will require further validation. Biomarkers that use MRI, PET, or both, to infer the presence of TDP-43 [4, 9, 27, 59] and alpha-synuclein [17, 21, 48, 49, 58, 62] have been proposed, but still have only limited validation. Therefore, understanding how distinct neurodegenerative pathologies impact specific subregions of the MTL may help inform detection of AD and other neuropathologies in vivo and monitoring of patterns of pathology-specific downstream neurodegeneration in clinical research.

Postmortem work combining neuroimaging with pathology measurements is frequently used to study these associations. Prior studies of pathology-atrophy associations in the MTL, using either ex vivo or antemortem MRI, have shown subtle but somewhat inconsistent differences in the relationships between different pathologies and MTL subregional cortical thickness [19, 31, 53, 77, 96]. Such studies have primarily utilized staging systems or semi-quantitative ordinal ratings of pathology (e.g., none, mild, moderate, severe) derived at the time of neuropathological examination as measures of pathological burden. Staging methods, such as the Braak staging system for tau pathology in AD [6], are global descriptors of topographical spread and pathology progression through the brain, but offer only limited information on pathology burden in specific regions. Semi-quantitative visual ratings of pathology in individual brain regions provide an alternative, but are subjective and difficult to harmonize across studies, and cannot capture the variation of pathology burden within a given rating. Recent advancements in histological image analysis have allowed for the development of automated tools to quantify pathology burden from whole-slide digital images. A number of approaches for quantifying amyloid-beta plaques and tau neurofibrillary tangles (NFTs) in AD show high accuracy when compared to manual counts [71, 81, 84, 90, 99]; others have also been developed to quantify TDP-43 pathology [56]. The limited number of studies that used quantitative pathology measures in neurodegenerative disease studies found correlations with clinical features such as cognitive testing and duration of illness [16, 24, 47, 91] that were not found when using classic staging schemes or semi-quantitative ordinal ratings [37, 61, 92].

Here, we aim to characterize the patterns of MTL neurodegeneration, as measured by antemortem MRI, associated with two proteinopathies that profoundly impact MTL atrophy: phosphorylated tau (p-tau) pathology in AD neuropathological change (ADNC), and phosphorylated TDP-43 (pTDP-43) pathology in LATE neuropathological change (LATE-NC). Both pathologies follow stereotypical progression patterns within the MTL [5, 6, 45, 69] and are associated with MTL atrophy and cognitive decline [3, 18, 29, 47, 70, 85]. According to Braak & Braak, tau pathology in AD begins in the transentorhinal region before spreading to hippocampus and eventually the isocortex [5, 6]. Early stages of LATE-NC involve TDP-43 pathology in the amygdala before progression to the hippocampus and finally neocortex [45, 69]. Tau and TDP-43 also have high rates of co-occurrence in the MTL, and their co-occurrence is linked to higher odds of dementia and more rapid disease progression [39, 43, 50]. Thus, disentangling the relationship of MTL atrophy with these two pathologies has important implications for patient care and clinical trials. Here, we evaluate the association of p-tau and pTDP-43 pathology with structural measures of MTL subregions, using antemortem neuroimaging of the MTL to promote generalizability of our findings to in vivo imaging biomarkers. Based on previous studies of these pathologies and structure, we hypothesize that tau and TDP-43 pathology will be linked to different patterns of MTL atrophy, with greatest p-tau associations in the transentorhinal region (medial BA35, lateral ERC) and throughout the long axis of the hippocampus [3, 77, 85, 93, 96] while pTDP-43 will be more tightly linked to atrophy in the anterior hippocampus and ERC [2, 27, 77, 95]. An additional aim was to compare the utility of quantitative measures of p-tau and pTDP-43 burden extracted from digital histology images using a machine learning approach [22, 103] to semi-quantitative ratings assigned by neuropathologists in modeling structure. We hypothesize that quantitative measures will be more strongly associated with MTL atrophy compared to semi-quantitative ratings, particularly when restricting the analysis to individuals in the most severe stages of disease, where we expect ordinal semi-quantitative ratings to experience a “ceiling effect,” limiting their ability to explain variation in MTL atrophy. We test these hypotheses in a large cohort of brain donors (n = 140) with antemortem MRI and postmortem neuropathology examination.

Materials and methods

Deceased research study participants from the Penn Alzheimer’s Disease Research Center and Penn Frontotemporal Dementia Center who underwent antemortem MRI at Penn, donated their brain for research, and underwent autopsy at the Penn Center for Neurodegenerative Disease Research (CNDR) were included in this study. Patients whose neuropathological diagnoses included non-AD tauopathies, frontotemporal lobar degeneration (FTLD), or amyotrophic lateral sclerosis (ALS) were excluded, resulting in patients with diagnoses including ADNC, primary age-related tauopathy (PART), LBD, cerebrovascular disease (CVD), cerebral amyloid angiopathy (CAA), HS, and LATE-NC. We refer to this cohort composition as the “Alzheimer’s continuum” since it includes the full range of tangle and plaque pathology as well as concomitant pathologies that are common in ADNC.

Antemortem MRI acquisition and analysis

All patients underwent T1-weighted antemortem imaging during life at the University of Pennsylvania. All imaging procedures during life were performed with informed consent in accordance with Penn Institutional Review Board Guidelines and the Declaration of Helsinki. These scans took place from 1999 to 2021 at 1.5 T or 3 T magnetic field strengths. Magnetic field strength, scanner type, and acquisition parameters (i.e., repetition time, echo time, flip angle, and voxel output size) varied over the course of the study; additional information on these values can be found in the Supplementary Information. Scans were processed using T1-ASHS, a previously validated pipeline for automatic segmentation of MTL subregions [101]. Segmented regions include anterior hippocampus (AH), posterior hippocampus (PH), entorhinal cortex (ERC), Brodmann areas 35 (BA35) and 36 (BA36), which are parts of the perirhinal cortex, and parahippocampal cortex (PHC), as well as the surrounding white matter, sulci, and meninges. A surface-based pipeline CRASHS [105] based on the Nighres package [35] was used to extract the mid-surface between the gray/white matter and gray matter/CSF boundaries, inflate this surface to achieve a simplified, smoothed shape, and perform surface-based registration between inflated surfaces and a surface template. Gray matter thickness was computed for each participant and mapped into the template, allowing thickness measures from all participants to be analyzed at each point of the template. Additionally, summary measures for each subregion were computed, following [101]: median thickness measurements for the more “sheet-like” MTL extrahippocampal subregions (ERC, BA35, BA36, and PHC) and total volume for AH and PH. All segmentations were visually inspected (see [94] for the protocol); poor-quality scans or failed segmentations were excluded from analysis on a per hemisphere basis. For all participants, the scan closest to death with the most hemispheres passing segmentation quality control was utilized. Only scans within 10 years before death were included in the final imaging analyses.

Neuropathological staging (semi-quantitative measures)

For each donor, one hemisphere underwent extensive pathology processing at the Penn CNDR, as described previously [88]. The hemisphere side to be evaluated was generally randomly selected. Tissue was embedded in paraffin blocks and cut into 6 µm sections for immunohistochemistry using primary antibodies of NAB228 (monoclonal antibody [mAb], 1:30,000, generated in the CNDR) to detect amyloid deposits and for Thal staging, p-tau PHF1 (mAb, 1:3000, a gift from Dr. Peter Davies) to detect p-tau deposits and for Braak staging, 1D3 (mAb, 1:300, a gift from Dr. Manuela Neumann and Dr. Elisabeth Kremmer) to detect pTDP-43 deposits, and Syn303 (mAb, 1:30,000, generated in the CNDR) to detect the presence of pathological ɑ-synuclein. Ratings of neuropathology in each region, as well as neuronal loss, were scored as absent (0), rare (0.5), mild (1), moderate (2), and severe (3). The semi-quantitative ratings of pathology are consensus-based ratings; each case is examined by an initial rater and then reviewed with an expert neuropathologist. Semi-quantitative evaluation of pathology is standard practice and has been previously validated in multicenter studies [64] and at Penn [38]. Neuropathologists also applied the 2012 NIA-AA criteria for global neuropathological Thal, Braak, and Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) staging, as well as neuropathological diagnoses [65]. LATE-NC was staged according to established criteria [69] using semi-quantitative neuropathology ratings of pTDP-43 pathology. Ordinal ratings of pathology in the amygdala were considered for stage 1, in the dentate gyrus, cornu ammonis 1 (CA1) and subiculum, and entorhinal cortex for stage 2, and in the middle frontal gyrus for stage 3. Entorhinal cortex ratings were most frequently at the level of the hippocampal body though occasionally at the amygdala, and as such were used to determine LATE-NC stage 2 pathology. A rating of 1 (mild) or greater in these regions was considered positive. Finally, semi-quantitative ratings of pathology in three MTL regions (entorhinal cortex, CA1/subiculum, and dentate gyrus) were averaged to create semi-quantitative ratings of p-tau and pTDP-43 pathology in the MTL; higher values of this average correspond to more pathology. This process was repeated for ratings of neuronal loss. These averages are referred to as the semi-quantitative MTL p-tau (or pTDP-43 or neuronal loss) rating throughout.

Quantitative pathology methods

Histology slides previously used for neuropathological staging were digitally scanned on the Huron LT120 optical scanner at 0.4 × 0.4 µm resolution or on the 3DHistech Panoramic P250 scanner at 0.24 µm × 0.24 µm resolution and transferred to an open-source web application, “PICSL histology annotation server (PHAS)” (https://github.com/pyushkevich/histoannot), for viewing and annotating. A deep learning based weakly supervised learning (WSL) algorithm, WildCat [22], was adapted to the task of labeling pathological inclusions in histology slides, similar to earlier work [103]. WildCat examines image patches of dimension 224 × 224 pixels (sampled from digital histology images resampled to 0.4 µm × 0.4 µm resolution) and assigns each patch a class label (e.g., tangles, threads, background, or other for p-tau). WildCat also generates a spatial attention heat map that indicates the location and intensity of features driving the classification decision. Training data for WildCat were generated by human raters who placed thousands of rectangular boxes around examples of different classes of pathological inclusions. In previous work, we trained WildCat to detect p-tau threads, p-tau tangles, normal-appearing tissue, and slide background on 50 µm thick AT8-stained slides and showed that the attention maps generated by the algorithm on novel slides unseen during training provide a rough pixel-level segmentation of tangles. Summary measures of tangle burden obtained by integrating these attention maps over regions of interest were strongly correlated with manual tangle counts in these regions, as well as with an expert’s semi-quantitative ratings [103].

In the current work, two new classifiers were trained (i) for labeling p-tau inclusions in PHF1-stained 6 µm thick sections and (ii) for labeling pTDP-43 inclusions in 1D3-stained 6 µm thick sections, as detailed below.

P-tau pathology training data

For p-tau, training data for WildCat was generated by 22 raters over the course of a single day event. Raters received training from and were supervised by neuropathology experts, who also participated in annotating training data. Raters had access to the full-resolution whole slide images of p-tau PHF1-stained slides on the PHAS system and the patients’ neuropathological diagnoses and global pathological burden ratings. Using the PHAS system, raters were asked to draw boxes to mark a variety of neuronal, dendritic, and glial p-tau inclusions as well as normal appearing tissue, artifacts, and background. Criteria for annotation were based on the training from [103], with additional information from [11, 25, 26, 54, 65, 89] and input from the expert neuropathologists. Sections from 37 specimens spanning different neuropathological diagnoses and stages were included for training. Further details on the training data and criteria are given in the Supplementary Information.

Training labels were grouped into four broader classes of background (including healthy neurons, tissue, non-tissue background, artifact, stain background, and dirt), tangles (including NFTs, non-pyramidal tangles, and pre-tangles), threads (including gray matter threads, neuritic plaques, tangle associated neuritic clusters, axonal threads, and grains), and other (including astrocytic p-tau, coiled bodies, ballooned neurons, and globular glial tauopathies) for training of WildCat. A total of 10,304 training examples were generated (Supplemental Table 1).

pTDP-43 pathology training data

For pTDP-43, training data for WildCat were generated by 7 raters over the course of three months. Neuropathology experts trained and supervised these raters, who had access to the full-resolution whole slide images of pTDP-43 1D3-stained slides but were blinded to any neuropathology ratings or diagnoses. Raters were asked to use boxes to mark a variety of neuronal, neuritic, and glial pTDP-43 pathologies, as well as normal appearing tissue. Criteria for the training were created through discussion with expert neuropathologists and review of current research [7, 8, 11, 44, 57, 68, 72, 73, 86]. Sections from 65 specimens across neuropathological diagnoses and stages were included for training (more information on the specimens and annotation criteria are provided in the Supplementary Information).

Training labels were grouped into classes of background (including healthy neurons, background, tissue, dirt, artifact, and vacuoles), neuronal/glial (round neuronal cytoplasmic inclusions (NCIs), ring-like NCIs, neuronal intranuclear inclusions (NIIs), and white matter oligodendroglial inclusions), neuritic (short dystrophic neurites, long dystrophic neurites, white matter threads, and grains), and non-specific staining (including granulovacuolar degeneration, neuromelanin, and other non-specific staining). A total of 12,489 training examples were generated (Supplemental Table 2).

WildCat training

To evaluate the accuracy of WildCat patch-level classification, five-fold cross-validation experiments were conducted. Annotated specimens for each pathology were randomly split into five subsets, and, in each cross-validation experiment, four subsets were used for WildCat training and validation (a small validation subset is used to determine stopping criteria to avoid overfitting) and the remaining subset was used as the test set. For each cross-validation experiment, we quantified the sensitivity and specificity of the labels predicted by the corresponding WildCat model. Finally, for each pathology, a single WildCat model was trained using patches from all the annotated specimens. These models were used in subsequent analyses. A visualization of the WildCat pipeline and example patches for both p-tau and pTDP-43 are shown in Fig. 1.Fig. 1 WildCat pipeline and sample activation maps. a Pipeline for WildCat classification of p-tau inclusions. The pipeline is the same for the WildCat trained for pTDP-43 inclusions, replacing tangles and threads with neuronal/glial and neuritic pTDP-43. b Sample patch level activations for classification of p-tau inclusions. Shown are example patches and heatmaps showing the activation for each class (tangles, threads, other, and background). c Sample patch level activations for classification of pTDP-43 inclusions. Shown are example patches and heatmaps showing the activation for each class (neuronal/glial pTDP-43, neuritic pTDP-43, non-specific staining, and background). Green areas in (b) and (c) correspond to areas of higher activation

Anatomical sampling regions of interest

To imbue quantitative measurements of pathology with anatomic specificity, sampling boxes in anatomical regions of interest (ROIs) were annotated on PHF1 and 1D3 digital histology images from all cases with eligible neuropathological diagnoses and any antemortem imaging. Given the large amount of effort that such annotation requires, only histology sections including the hippocampus were annotated in the current paper, as it is an early site of both p-tau and pTDP-43 pathology. The sampling boxes were placed using the PHAS system. Anatomical regions labeled included CA1, CA2, and CA3, the granular cell layer of the dentate gyrus, the hilus of the dentate gyrus, a transition area between CA1 and subiculum, a transition area between subiculum and presubiculum, three areas within the entorhinal cortex corresponding to the areas most medial, central, and lateral on the slide, BA35, and BA36; see the Supplementary Information for details on the annotation protocol.

Annotation yielded a total of 1,585 regions of interest annotated across 200 PHF1 slides. To reduce manual effort, piecewise diffeomorphic deformable registration between hippocampal PHF1 and 1D3 slides was performed for each individual using the multi-stain registration algorithm in [73] and ROIs were mapped onto the corresponding 1D3 slides. ROIs were manually corrected for registration errors and to avoid areas of significant tissue damage. When registration failed or was not possible (e.g., p-tau and pTDP-43 slides in different blocks, or only one stain available), the pTDP-43 slide was manually annotated following the same protocol. Registration and manual annotation resulted in 1,576 regions of interest annotated across 194 1D3-stained slides. Examples of annotated p-tau and pTDP-43 sections are shown in Fig. 2.Fig. 2 Examples of annotated hippocampus histology sections and corresponding WildCat activation maps. a A p-tau-stained section (left) and a pTDP-43-stained section (right) with sampling regions of interest annotated. The p-tau section is from a case with high ADNC; the pTDP-43 section is from a case with LATE-NC stage 2. The colors of each label are given to the right. b Shown is the EC-lat (lateral area of entorhinal cortex) region of interest labeled from the same section, the WildCat activation heat map for tangles and threads (p-tau) or neuronal/glial and neuritic pTDP-43, and the same ROI broken into approximately equal size contiguous segments from which the average activation measure is taken. These segments are obtained using the METIS graph partitioning algorithm [51]. Yellow and green areas indicate areas of higher activation. CA cornu ammonis, CA1-SUB cornu ammonis 1/subiculum, DG-H hilus of the dentate gyrus, DG-GCL granular cell layer of the dentate gyrus, SUB-PrS subiculum/presubiculum, EC-med medial portion of the entorhinal cortex visible on the section, EC-ctr central portion of entorhinal cortex visible on the section, EC-lat lateral portion of entorhinal cortex visible on the section, BA Brodmann area

Extraction and evaluation of ROI-level quantitative pathology measures

Heat maps of pathology burden (tangles and threads for p-tau; neuronal/glial and neuritic for pTDP-43) were generated for each sampling box by applying the corresponding trained WildCat model. From each of these heat maps, a summary measure of pathology burden was generated by breaking each sampling box into small (~ 200 × 200 µm) segments (examples of these segments are shown in Fig. 2), averaging the heat map over each segment, and taking a summary statistic (e.g., 90th percentile, mean) across all segments. To validate these box-level summary measures, hippocampal sections from cases with passing quality antemortem MRI not included in the model training were stratified by the semi-quantitative MTL ratings of p-tau, and 64 sampling boxes were randomly selected for validation, and likewise for pTDP-43. An expert rater (p-tau: SA, pTDP-43: JR) examined each sampling box and assigned separate semi-quantitative ratings (0–3) of tangle and thread pathology for p-tau or neuronal/glial and neuritic pathology for pTDP-43. As the 64 sampling boxes for pTDP-43 showed a skew towards less pathology, 29 additional sampling boxes were selected from the high MTL pTDP-43 strata to ensure the full range of pathology was captured for validation. Quantitative measures for each pathology were generated for each sampling box in this validation dataset. The agreement between the experts’ ratings and the corresponding box-level quantitative measures generated from aggregating the segment-level averages was evaluated statistically via receiver operating characteristic (ROC) curve analysis and Mann–Whitney tests. For each of the four types of pathology measures, we tested the ability of the box-level summary quantitative ratings to discriminate between adjacent semi-quantitative categories. The analysis was repeated using different summary statistics (mean, median, maximum, 25th, 75th, 90th, 95th, and 99th quantiles) to determine the summary statistic that most consistently agreed with experts’ ratings across all four pathology categories. Additional details can be found in the Supplementary Information.

For subsequent analyses, quantitative pathology measures were calculated for each anatomical ROI rather than each sampling box since most anatomical ROIs were sampled using several boxes. To do this, segment-level heat map averages for each anatomical ROI were pooled across the multiple sampling boxes, and the best-performing quantile statistic was applied. This yielded quantitative measures of tangle and thread burden for each anatomical ROI in each p-tau section, and, similarly, measures of neuronal/glial and neuritic burden for each anatomical ROI in each pTDP-43 section.

Hippocampus-level summary quantitative pathology measures

Summary measures of hippocampal pathology burden were generated for each participant by averaging the ROI-level pathology measures across CA1, CA2, CA3, CA1/subiculum, subiculum/presubiculum, the hilus of the dentate gyrus, and the granular cell layer of the dentate gyrus. We included these regions and not others in the average because they could be commonly sampled across our dataset (i.e., not frequently torn or damaged) and because they can be identified consistently along the hippocampal longitudinal axis. Conversely, ROIs from the MTL cortex were annotated, but not included in the average, because there was significant variation in CNDR slides as to where along the longitudinal axis the hippocampus was sectioned (body vs. head). Without a reference ex vivo MRI or serial histology, it is difficult to estimate the exact location of a histology slide along the longitudinal axis. This variation makes it difficult to assess whether the MTL cortex present on each slide contains entorhinal cortex and BA35/36, as expected on more anterior sections, or parahippocampal cortex, as expected on more posterior sections. A summary hippocampal pathology measure was not computed for a participant if fewer than three anatomical ROIs from the list above could be annotated; ROI measurements available in both dentate gyrus subregions were considered as only one region towards this count (though their measures were calculated separately, and both included in the average).

To make the scale of the hippocampus-level summary quantitative pathology measures more meaningful, we applied a linear transformation to the hippocampal quantitative summary measures based on the linear fit between the untransformed quantitative measures and the corresponding semi-quantitative MTL p-tau or pTDP-43 ratings. This transformation puts the summary quantitative measures roughly in a range between 0 and 3, with 0 corresponding to little or no burden and 3 corresponding to severe burden. This simple linear transformation has no impact on the associations with MRI-derived measures and is only performed to improve the interpretability of these measures. We refer to this transformed average as the “summary measure” for each inclusion type (e.g., p-tau tangles) throughout.

Statistics

All analyses were performed in R version 4.3.0. Validation of the quantitative pathology measurements in individual sampling regions was evaluated with one-sided Mann–Whitney area under the curve (AUC) tests between adjacent expert ratings. Additional validation of the quantitative summary measures included Spearman correlations of the summary measure with the average MTL p-tau and pTDP-43 ratings and average neuronal loss rating in the MTL. Quantitative pathology summary scores were also compared to Braak (p-tau) and LATE-NC (pTDP-43) staging through one-sided Mann–Whitney AUC tests. These pathology analyses included all cases with p-tau or pTDP-43 quantitative summary pathology measurements regardless of MRI or segmentation quality.

For the pathology and imaging analyses, cases with quantitative p-tau and pTDP-43 summary measures were included if the antemortem MRI passed segmentation and image quality control in the hemisphere ipsilateral to the one used for neuropathology. This group is referred to as the “whole imaging cohort.” To examine the relationships between MRI-derived measures and quantitative pathology summary measures in cases with more severe pathology burden, where we hypothesized that we would observe “ceiling effects” with semi-quantitative measures, we defined the advanced ADNC subgroup as participants with Braak stages V or VI. As only a small number of individuals had LATE-NC stage 2 or 3, we defined the LATE-NC subgroup as participants with LATE-NC stage 1 or greater. In the overall cohort and these two subgroups, linear models were fit for each MTL subregion using T1-ASHS derived volume (AH/PH) or median thickness (ERC/BA35/BA36/PHC) as the dependent variable and p-tau and pTDP-43 summary pathology measures as the independent variables.

To determine which combination of pathology predictors to analyze in these models (i.e., semi-quantitative or quantitative measures for both p-tau and pTDP-43), we performed post-hoc analyses to evaluate the differences in the strength of the association between quantitative summary pathology measures and semi-quantitative ratings with structure. For p-tau, we generated two linear models for each imaging ROI; a quantitative model including the tangles and threads summary measures and covariates of age, antemortem interval (time from MRI to date of death), sex, MRI field strength, and the semi-quantitative MTL pTDP-43 rating, as well as intracranial volume (ICV) for the 2 regions with volume computed (AH/PH). The semi-quantitative model included the MTL p-tau rating and the same covariates. These models included all patients with ipsilateral imaging and quantitative p-tau summary measures available. We computed the differences between these two models in adjusted R2, prediction error, Akaike’s information criterion with correction for small sample sizes (AICc), and Bayesian information criterion (BIC). Prediction error was calculated as the sum of the absolute value (L1 norm) of the residuals for each model. A difference of greater than 2 in BIC was considered positive evidence of a better model fit, greater than 6 as strong evidence, and greater than 10 as very strong evidence [75]. These tests were also performed for pTDP-43 with pathology variables of hippocampal neuronal/glial and neuritic pTDP-43 summary measures in the quantitative model, the MTL pTDP-43 rating in the semi-quantitative model, and the semi-quantitative MTL p-tau rating as a covariate for both.

Based on these model comparisons, we selected the best pathology measures (either quantitative or semi-quantitative measures for p-tau and pTDP-43) to analyze in one linear model for each pathology to structure analysis in both the whole imaging cohort and the two subgroups (advanced ADNC and LATE-NC). Cases were included if the antemortem MRI passed segmentation quality control in the hemisphere ipsilateral to the one used for neuropathology and they had the required pathology measurements. Each model included age at death, sex, MRI field strength, and antemortem interval as nuisance covariates. ICV, computed using [101], was included as an additional covariate for AH and PH volume, since hippocampal volume measures, but not cortical thickness measures, are expected to correlate with head size [1, 80]. For each fitted model, the standardized β coefficients, t-statistics, and one-sided p-values for the contrasts of interest (e.g., association between tangle summary measure and ERC median thickness) were computed and reported. One-sided tests examined only negative associations as we expected all pathology to be negatively associated with neurodegeneration. We did not correct for multiple comparisons due to strong a priori hypotheses that structure and pathology should be negatively associated. However, results that would survive Bonferroni correction at P < 8.33e-03 (0.05 / 6 regions examined) are also indicated in the table.

Finally, three-dimensional surface maps were generated comparing the measure of gray matter thickness for each point in the MTL to the given pathology measures. These models were analogous to the ROI-level linear models, but with additional correction for multiple comparison using the cluster-level inference approach based on permutation testing [32], as implemented in the meshglm program (https://github.com/pyushkevich/cmrep). Specifically, after mapping all thickness values to a common geometrical template, fitting a linear model at each point in the template, and computing a t-statistic map for the contrast of interest at each point, we extracted contiguous clusters based on the threshold t > 2.0 and recorded the surface area of these clusters. We then computed the null distribution of maximum cluster surface area under random permutation (n = 10,000) of data labels, following the two-step Freedman-Lane procedure [28]. For each cluster, the one-sided corrected p value was computed as the proportion of permutation experiments in which a cluster of larger surface area was extracted.

Results

Demographics

229 brain donors met our inclusion criteria based on neuropathological diagnoses along the AD continuum and antemortem imaging. One autopsy was excluded due to a very young age at death (< 35 years). After excluding cases with fewer than three annotated anatomical ROIs, 196 cases with PHF1-stained p-tau sections were available in the p-tau pathology analyses (“Tau Pathology Cohort”) and 190 cases for pTDP-43 analyses (“TDP-43 Pathology Cohort”). For the imaging analyses, 163 patients had a usable antemortem T1-weighted MRI within 10 years prior to death that passed segmentation quality control on the ipsilateral hemisphere that was sectioned for neuropathology (“Imaging Cohort”); this number includes patients regardless of quantitative pathology availability. Demographics for these three groups are given in Table 1.Table 1 Demographics given for cases with quantitative p-tau data, quantitative pTDP-43 data, and ipsilateral MRI and segmentation passing QC

Cases with quantitative p-tau summary measures were included in the p-tau pathology-only analyses, and likewise for pTDP-43. The imaging cohort includes cases with usable antemortem MRI within 10 years prior to death and passing quality segmentation on the ipsilateral hemisphere to the one sectioned for neuropathology, regardless of the availability of quantitative pathology measures. A Amyloid, B Braak, C CERAD, LATE-NC Limbic predominant age-related TDP-43 encephalopathy neuropathologic change

Quantitative pathology validation

The average accuracy of five-fold patch-level cross-validation of WildCat was 85.2% for p-tau tangles and 73.5% for p-tau threads. Accuracy was higher for pTDP-43: 92.7% for neuronal/glial and 95.3% for neuritic inclusions (Table 2). Cross-validation results for each kind of inclusion within each of these classes are shown in Supplementary Tables 1 (p-tau) and 2 (pTDP-43). Low accuracy in the threads measurement is mostly attributable to misclassification of neuritic plaques (often as tangles or other) and axonal threads (most often as background), the latter of which is expected to be rare in our dataset as our sampling regions are placed only in gray matter. Astrocytic p-tau was also commonly misclassified in cross-validation as threads (~ 29%) rather than the correct “other” class.Table 2 Five-fold cross-validation for WildCat training, grouped by class, for p-tau and pTDP-43 measurements

Given is the number of patches (N) in each WildCat prediction class stratified by the true label annotated during training. The percentages evaluate the number of patches predicted in each class by WildCat within a given true label. Darker green colors indicate higher percentages

We tested multiple different summary statistics to generate quantitative pathology measures in individual sampling boxes, including various percentiles, median, and maximum, by comparing these statistics to semi-quantitative ratings of the same boxes by an expert rater. Different measures showed varying discrimination across pathological inclusion types. The 99th percentile was ultimately chosen for all further analyses as it showed high discrimination across all inclusion types (Supplementary Table 3 and Supplementary Figs. 1 and 2). Validation of the quantitative measures using the 99th percentile in individual sampling regions is shown in Fig. 3. The quantitative measures generally showed high discrimination between adjacent categories of semi-quantitative ratings and increased as semi-quantitative ratings increased, but quantitative measures from adjacent semi-quantitative ratings commonly overlapped. Poor separation was seen between the moderate and severe categories for threads (AUC = 0.48). This validation also revealed that WildCat incorrectly classified mossy fibers within the dentate gyrus as threads. Therefore, the threads measurements from the hilus of the dentate gyrus were excluded from all summary measures.Fig. 3 Validation of single sampling region pathology measures with expert ratings. Quantitative pathology measures in individual sampling regions were grouped by the ordinal semi-quantitative rating assigned by a visual read of the same sampling region by an expert (S.A. for p-tau, J.R. for pTDP-43). Each panel represents a different pathology type: tangles (a), threads (b), neuronal/glial pTDP-43 (c), and neuritic pTDP-43 (d). The same set of sampling regions were evaluated for tangles and threads, and similarly for the pTDP-43 measures. One-sided AUC and Mann–Whitney U-tests are shown between adjacent ratings (*p < 0.05, **p < 0.01, ***p < 0.001). CA cornu ammonis, CA1-SUB cornu ammonis 1/subiculum, SUB/PrS subiculum/presubiculum, DG-H hilus of the dentate gyrus, DG-GCL granular cell layer of the dentate gyrus, EC-lat lateral portion of entorhinal cortex visible on the section, EC-ctr central portion of entorhinal cortex visible on the section, EC-med medial portion of the entorhinal cortex visible on the section, BA Brodmann area

Pathology analyses

Quantitative pathology burden measures in each region of interest were plotted across Braak (Supplementary Fig. 3) and LATE-NC stages (Supplementary Fig. 4). Tangle and thread pathology generally increased in each region across Braak stages; this pattern was particularly clear for tangles. The pattern for pTDP-43 pathology across Braak stages was less consistent and varied by region and inclusion type. The high levels of pTDP-43 pathology, particularly neuritic, in Braak stage 0 patients is likely due to the small sample size (N = 4) and driven by one case with stage 2 LATE-NC. Braak stage V/VI cases tended to show higher levels of pTDP-43 pathology than other stages. When examining the LATE-NC staging, p-tau measures often showed slight increases across stages except for LATE-NC stage 3 cases (though note the small sample size for stage 3, N = 4). pTDP-43 measures generally showed similarly low levels of pathology for LATE-NC stages 0 and 1 and increasingly higher pathology burden in stages 2 and 3.

Both the quantitative pathology summary measures and the semi-quantitative ratings steadily increased across Braak (p-tau) and LATE-NC (pTDP-43) stages and showed varying degrees of separation between stages (Fig. 4). High variation was present in all measures within each Braak stage other than stage 0. The best discriminator between LATE-NC stages was the semi-quantitative MTL pTDP-43 ratings, as expected, as the individual semi-quantitative ratings in this average are also used to define LATE-NC stages. High variation was present within all quantitative summary pTDP-43 measures, including at low stages of LATE-NC. pTDP-43 measures showed low discrimination between stages 0 and 1 and between stages 2 and 3, but higher discrimination between stages 1 and 2.Fig. 4 Pathology measures compared to staging systems. a Quantitative summary measures and semi-quantitative ratings of p-tau pathology plotted across Braak stages for all patients with quantitative tau summary measurements (N = 196). b Quantitative summary measures and semi-quantitative ratings of pTDP-43 pathology plotted across LATE-NC stages for all cases with quantitative pTDP-43 summary measurements (N = 190). One-sided AUC and Mann–Whitney U-tests are shown between adjacent Braak and LATE-NC stages for each plot (*p < 0.05, **p < 0.01, ***p < 0.001). Tangles (Q) quantitative tangles summary measurement, Threads (Q) quantitative threads summary measurement, MTL p-Tau (SQ) semi-quantitative MTL p-tau rating, Neuronal/Glial pTDP-43 (Q) quantitative neuronal/glial pTDP-43 summary measurement, Neuritic pTDP-43 (Q) quantitative neuritic pTDP-43 summary measurement, MTL pTDP-43 (SQ) semi-quantitative MTL pTDP-43 rating

Quantitative pathology summary measures showed varying strength but significant correlations with semi-quantitative ratings of the same MTL pathology; neuronal measures (tangles: R = 0.69, p < 2.2 × 10–16, neuronal/glial pTDP-43: R = 0.53, p = 3.5 × 10–15) showed stronger correlations than neuritic measures (threads: R = 0.39, p = 1.9 × 10–8, neuritic pTDP-43: R = 0.40, p = 7.8 × 10–9) (Fig. 5). Correlations with the average neuronal loss in the MTL followed a similar pattern, though weaker, and the threads summary measure did not significantly correlate with neuronal loss (Supplemental Fig. 5).Fig. 5 Scatterplots of summary pathology measures and semi-quantitative ratings of the same pathology in the MTL. a Scatterplots of the quantitative tangle and threads summary measures plotted against the semi-quantitative MTL p-tau rating (N = 196). b Scatterplots of the quantitative neuronal/glial and neuritic pTDP-43 summary measures plotted against the semi-quantitative MTL pTDP-43 rating (N = 190). Spearman correlations between each summary measure and its corresponding semi-quantitative rating are shown in the top left of each plot. Tangles (Q) quantitative tangles summary measurement, Threads (Q) quantitative threads summary measurement, Neuronal/Glial pTDP-43 (Q) quantitative neuronal/glial pTDP-43 summary measurement, Neuritic pTDP-43 (Q) quantitative neuritic pTDP-43 summary measurement

Pathology and imaging

Scatterplots between the six MRI ROI-level volume/thickness measures and available ipsilateral quantitative and semi-quantitative pathology measures are plotted in Fig. 6 for the 126 patients for whom all these measures are available (Supplemental Table 4 for demographics). A strong relationship between most MRI measures and the quantitative p-tau tangles summary measure was observed, likely largely driven by individuals with advanced AD (Braak stage V/VI, N = 75, labeled with a cross). These individuals exhibited a greater range of variation in the quantitative tangles measure than in the semi-quantitative p-tau measure. All pTDP-43 measures are highly skewed towards zero, reflecting the lack of hippocampal pTDP-43 pathology in most participants. A relationship was also observed between most MRI measures and both the semi-quantitative pTDP-43 rating and the neuronal/glial quantitative pTDP-43 summary measure in the participants with LATE-NC stages 1–3 (n = 24, red color).Fig. 6 Scatterplots of tau and TDP-43 pathology measures compared to ipsilateral cortical thickness and volume measurements. The sample includes all cases with both quantitative p-tau and pTDP-43 summary pathology measures available and ipsilateral cortical thickness and volume data (N = 126). Cases with LATE-NC stages 1–3 (N = 24) are colored in red. Cases with Braak stages of V/VI (N = 75) are represented by a cross. A regression line without any covariates is also included in each plot. AH (vol) anterior hippocampus volume (mm3), PH (vol) posterior hippocampus volume (mm3), ERC (thk) entorhinal cortex median thickness (mm), BA35 (thk) Brodmann area 35 median thickness (mm), BA36 (thk) Brodmann area 36 median thickness (mm), PHC (thk) parahippocampal cortex median thickness (mm), Tangles (Q) quantitative tangles summary measurement, Threads (Q) quantitative threads summary measurement, MTL p-Tau (SQ) semi-quantitative MTL p-tau rating, Neuronal/Glial pTDP-43 (Q) quantitative neuronal/glial pTDP-43 summary measurement, Neuritic pTDP-43 (Q) quantitative neuritic pTDP-43 summary measurement, MTL pTDP-43 (SQ) semi-quantitative MTL pTDP-43 rating

Model comparisons

We statistically compared models that included quantitative vs. semi-quantitative measures of p-tau and pTDP-43 burden (Table 3). For each MTL subregion, Table 3 lists the differences in Radj2, prediction error (L1 norm of the residuals), AICc, and BIC between the two non-nested models fitted using the same dependent variable (MRI measures) and nuisance covariates and differing by the type of pathology measures used as the independent variable (quantitative or semi-quantitative). For the p-tau model comparison, we used all cases that had antemortem imaging of the ipsilateral hemisphere and quantitative hippocampal p-tau summary pathology measurements (N = 140, Supplemental Table 4 for demographics). The model with quantitative p-tau summary measures had a higher Radj2 and lower prediction error, AICc, and BIC value in every region than the model with semi-quantitative p-tau ratings, except for BIC in the association with PHC. We had positive evidence (∆BIC > 2) that the quantitative model fit better for BA35 and very strong evidence (∆BIC > 10) for AH, PH, and ERC.Table 3 Comparisons of quantitative and semi-quantitative models for p-tau and pTDP-43

p-Tau	
	∆ Radj2	∆Prediction Error	∆AICc	∆BIC	
AH (volume)	0.084	− 2035.718	− 15.654	− 12.998	
PH (volume)	0.111	− 2356.001	− 25.381	− 22.724	
ERC (thickness)	0.110	− 1.529	− 18.455	− 15.764	
BA35 (thickness)	0.043	− 0.806	− 5.219	− 2.528	
BA36 (thickness)	0.027	− 0.572	− 2.829	− 0.138	
PHC (thickness)	0.017	− 0.306	− 1.414	1.277	
pTDP-43	
	Radj2	∆Prediction error	∆AICc	∆BIC	
AH (volume)	− 0.004	− 280.254	1.966	4.641	
PH (volume)	− 0.045	437.871	10.568	13.243	
ERC (thickness)	0.014	− 0.142	− 1.123	1.586	
BA35 (thickness)	0.032	− 0.582	− 3.777	− 1.068	
BA36 (thickness)	− 0.004	− 0.003	1.809	4.518	
PHC (thickness)	0.013	− 0.429	− 0.944	1.765	
For each ROI and pathology, the differences between the models using quantitative or semi-quantitative pathology in adjusted R2, in prediction error (calculated as L1 norm of the residuals), in AICc, and in BIC were calculated. Note that the large difference in range in prediction error between the hippocampal measures and extracortical regions is likely a result of using volume for AH/PH and thickness for the extracortical regions (ERC, BA35, BA36, PHC). A positive difference in Radj2 favors the quantitative model. A negative difference in prediction error, AICc, and BIC favors the quantitative model. AH anterior hippocampus volume, PH posterior hippocampus volume, ERC entorhinal cortex median thickness, BA35 Brodmann area 35 median thickness, BA36 Brodmann area 36 median thickness, PHC parahippocampal cortex median thickness. AICc Akaike information criterion with correction for small sample sizes, BIC Bayesian information criterion

The model comparisons for quantitative vs. semi-quantitative pTDP-43 (N = 142, Supplemental Table 4 for demographics) showed much less consistency across MTL regions. Quantitative models produced a higher Radj2 for three of the four cortical regions, but not either hippocampal region, and had a smaller prediction error for every region other than PH. AICc values favored quantitative pathology measures for ERC, BA35, and PHC, but these effects were weak or in the opposite direction when using BIC values. BIC values showed positive evidence (∆BIC > 2) favoring the semi-quantitative measures for BA36 and AH, and very strong evidence (∆BIC > 10) favoring the semi-quantitative model for PH.

Overall, the model comparison suggests that quantitative hippocampal p-tau summary measures provide a better fit to the MTL structural model than the semi-quantitative p-tau rating. However, for pTDP-43 there is no such clear advantage of using quantitative measures, and indeed, semi-quantitative rating may be preferable. Based on this conclusion, we include quantitative p-tau summary measurements and semi-quantitative MTL pTDP-43 ratings in the thickness-to-structure analyses in the following sections.

Pathology and structure analyses

140 patients had ipsilateral imaging and quantitative p-tau hippocampal summary pathology measurements available and were included in these analyses (Supplemental Table 4 for demographics). T-statistics from the linear models fitted in this entire cohort and in the advanced ADNC and LATE-NC subgroups are shown in Table 4. In the whole cohort, the tangles summary measures showed significant negative associations with all regions, with the strongest effects in absolute terms with ERC (β =  − 0.47). Significant associations were also found between the MTL pTDP-43 rating and smaller volumes and thickness in AH, PH, ERC, and BA36, but the β coefficients were smaller in absolute terms than the associations with the tangles measure. The strongest association of pTDP-43 was with PH (β =  − 0.30). In the LATE-NC subgroup (N = 31), the tangles summary measure showed significant negative associations with AH, PH, and ERC, and this association was strongest with ERC (β =  − 0.67). Additionally, significant associations were found between the MTL pTDP-43 rating and all regions except BA35 and PHC, and the strongest effect was with PH (β =  − 0.48), though AH and ERC also showed similarly high associations (β =  − 0.47 for both). We repeated these analyses in the LATE-NC stage 1 and 2/3 patients separately (Supplementary Table 5). Here, we found multiple significant associations with p-tau (AH, ERC, and BA35) but not with pTDP-43 in the LATE-NC stage 1 patients (N = 17) and no significant associations with either pathology in the LATE-NC stage 2/3 patients (N = 14). Finally, in the advanced ADNC subgroup (N = 85), the quantitative tangles summary measure was again significantly associated with smaller volumes and thickness in all subregions except PHC and BA36, with the strongest association with PH (β = − 0.51). Finally, the MTL pTDP-43 rating also showed significant associations with AH, PH, and ERC, which was strongest for PH (β = − 0.26).Table 4 Pathology to structure analyses in cases with quantitative p-tau pathology measures and ipsilateral MRI

	AH (vol)	PH (vol)	ERC (thk)	BA35 (thk)	BA36 (thk)	PHC (thk)	
Whole cohort (N = 140, df = 131/132)	
Tangles (Q)	β = − 0.30,	β = − 0.44,	β = − 0.47,	β = − 0.31,	β = − 0.23,	β = − 0.27,	
t = − 3.94,	t = − 6.37,	t = − 5.84,	t = − 3.41,	t = − 2.51,	t = − 3.01,	
p = 6.50e-05	p = 1.48e-09	p = 1.97e-08	p = 4.28e-04	p = 0.007	p = 0.002	
Threads (Q)	β = − 0.07,	β = − 0.01,	β = 0.01,	β = 0.09,	β = − 0.10,	β = − 0.01,	
t = − 0.88,	t = − 0.16,	t = 0.17,	t = 0.99,	t = − 1.14,	t = − 0.07,	
p = 0.190	p = 0.436	p = 0.566	p = 0.838	p = 0.129	p = 0.473	
MTL pTDP-43 (SQ)	β = − 0.28,	β = − 0.30,	β = − 0.26,	β = − 0.04,	β = − 0.16,	β = − 0.08,	
t = − 3.91,	t = − 4.73,	t = − 3.46,	t = − 0.50,	t = − 1.92,	t = − 0.98,	
p = 7.29e-05	p = 2.86e-06	p = 3.58e-04	p = 0.308	p = 0.028	p = 0.164	
LATE-NC Cohort (N = 31, df = 22/23)	
Tangles (Q)	β = − 0.47,	β = − 0.37,	β = − 0.67,	 β = − 0.35,	 β = − 0.14,	 β = − 0.26,	
t = − 3.01,	t = − 1.82,	t = − 3.79,	t = − 1.38,	t = − 0.59,	t = − 1.22,	
p = 0.003	p = 0.041	p = 4.72e-04	p = 0.091	p = 0.281	p = 0.118	
Threads (Q)	β = 0.33,	β = 0.14,	β = 0.26,	β = 0.21,	β = 0.07,	β = 0.08,	
t = 2.42,	t = 0.79,	t = 1.64,	t = 0.92,	t = 0.33,	t = 0.44,	
p = 0.988	p = 0.782	p = 0.942	p = 0.818	p = 0.629	p = 0.668	
MTL pTDP-43 (SQ)	β = − 0.47,	β = − 0.48,	β = − 0.47,	 β = − 0.12,	 β = − 0.37,	 β = − 0.18,	
t = − 3.58,	t = − 2.76,	t = − 3.16,	t = − 0.55,	t = − 1.77,	t = − 1.02,	
p = 8.41e-04	p = 0.006	p = 0.002	p = 0.293	p = 0.045	p = 0.160	
Advanced ADNC cohort (N = 85, df = 76/77)	
Tangles (Q)	β = − 0.37,	β = − 0.51,	β = − 0.45,	 β = − 0.24,	 β = − 0.18,	 β = − 0.17,	
t = − 3.88,	t = − 6.27,	t = − 4.37,	t = − 2.11,	t = − 1.63,	t = − 1.44,	
p = 1.09e-04	p = 9.85e-09	p = 1.89e-05	p = 0.019	p = 0.054	p = 0.078	
Threads (Q)	β = − 0.13,	β = − 0.02,	β = − 0.06,	β = 0.03,	β = − 0.18,	β = − 0.02,	
t = − 1.38,	t = − 0.26,	t = − 0.61,	t = 0.26,	t = − 1.63,	t = − 0.15,	
p = 0.085	p = 0.399	p = 0.270	p = 0.602	p = 0.054	p = 0.442	
MTL pTDP-43 (SQ)	β = − 0.21,	β = − 0.26,	β = − 0.22,	β = 0.08,	β = − 0.04,	β = − 0.02,	
t = − 2.22,	t = − 3.07,	t = − 2.11,	t = 0.71,	t = − 0.32,	t = − 0.15,	
p = 0.015	p = 0.001	p = 0.019	p = 0.759	p = 0.375	p = 0.440	
The standardized β coefficient, t statistic, and p value for the association of each pathology measure with each imaging ROI is given within each cohort (top: whole cohort, N = 140, middle: subset of cases with LATE-NC stages 1–3, N = 31, bottom: subset of cases with Braak Stage V/VI, N = 85). Degrees of freedom (df) are also given for each cohort; the first number is for AH/PH associations due to the additional covariate of ICV, while the second number is for ERC, BA35, BA36, and PHC associations. Associations that would survive Bonferroni correction (P < 8.33e-03) are bolded; associations that are significant at P < 0.05 but would not survive corrections are italicized. AH (vol) anterior hippocampus volume, PH (vol) posterior hippocampus volume, ERC (thk) entorhinal cortex median thickness, BA35 (thk) Brodmann area 35 median thickness, BA36 (thk) Brodmann area 36 median thickness, PHC (thk) parahippocampal cortex median thickness, Tangles (Q) quantitative tangles summary measure, Threads (Q) quantitative threads summary measure, MTL pTDP-43 (SQ) semi-quantitative MTL pTDP43 rating

While model comparisons showed quantitative p-tau and semi-quantitative pTDP-43 measures were the better models of thickness, for completeness, in Supplemental Tables 6, 7, and 8, we also show these pathology-structure associations in each group including the semi-quantitative MTL p-tau rating or quantitative pTDP-43 summary measures. Notably, in the Braak V/VI patients, weak associations with the semi-quantitative p-tau rating were only statistically significant in two regions, compared to the quantitative p-tau tangles summary measure which showed strong associations across the MTL. In Supplemental Table 9, we also repeated the models in cases with an antemortem interval of less than 3 years and found similar results of widespread, significant associations of tau tangles compared to more limited associations of MTL subregions with TDP-43.

Pointwise regional thickness associations

Pointwise regional thickness association maps showed similar patterns to the ROI-level linear models (Fig. 7). In the whole cohort, the tangles summary measure showed strong associations across almost the entirety of the hippocampus and much of the extrahippocampal regions. We also found strong associations between the pTDP-43 rating and thickness across the hippocampus and in the entorhinal cortex. In the LATE-NC stages 1–3 subgroup, the tangles summary measure was significantly associated with thickness in a small portion of the entorhinal cortex. The pTDP-43 rating showed significant associations in the hippocampus and entorhinal cortex. Finally, in the advanced ADNC subgroup, the tangles summary measure again showed significant associations in the hippocampus, entorhinal cortex, and BA35, and the pTDP-43 measure showed associations only in the hippocampus. Similar to the ROI-level analyses, we also computed these pointwise analyses in additional models including the quantitative pTDP-43 or semi-quantitative p-tau measures (Supplementary Fig. 6) and found similar results.Fig. 7 Maps showing the association between pointwise cortical thickness and the tangles summary measure, threads summary measure, and the MTL pTDP-43 semi-quantitative rating. Each row represents a different cohort; top: the whole cohort (N = 140), middle: the subset of patients with LATE-NC stages 1–3 (N = 31), and bottom: the subset of patients with Braak staging of V/VI (N = 85). Each model is covaried for age, antemortem interval, MRI field strength, and sex. The clusters outlined in black indicate regions where a significant association was found (p < 0.05), and the p-values are indicated with a line. The white lines represent the borders between subfields. To the right, a 3-D representation of the MTL with the subregions labeled is shown; the white matter sections were not analyzed. Warm colors indicate areas with more negative t-statistics. AH anterior hippocampus, PH posterior hippocampus, ERC entorhinal cortex, BA35 Brodmann area 35, BA36 Brodmann area 36, PHC parahippocampal cortex, WM white matter

Discussion

In the current study, we developed and validated algorithms to quantify postmortem p-tau and pTDP-43 pathology and evaluated the associations of these pathologies with MTL neurodegeneration. We show that quantitative pathology measurements show high accuracy across multiple validation experiments and discriminate well between stages of pathology. We found that quantitative p-tau and semi-quantitative pTDP-43 measures were the best performing models of antemortem structural measures. In 140 cases with antemortem neuroimaging, we investigated the association of these pathology measures with MTL subregional thickness and volume measures. In both the ROI-level linear models and pointwise regional thickness analyses, we find significant associations between the hippocampal p-tau tangles summary measure and all MTL subregions, with the strongest associations in absolute terms with ERC, PH, and AH. Associations with pTDP-43 pathology were generally limited to AH, PH, and ERC.

Our findings of widespread p-tau associations are consistent with prior studies using both antemortem and postmortem imaging and pathology measures [3, 20, 77, 96, 98]. Model comparisons in the whole cohort showed very strong evidence that the quantitative p-tau summary pathology measures were a better model of antemortem structural measures than the semi-quantitative p-tau ratings for AH, PH, and ERC, and showed at least weak evidence in favor of quantitative measures for all MTL subregions. Additionally, within cases with Braak stages of V/VI, the tangles summary measurement was significantly associated with subregions across the MTL, while the semi-quantitative MTL p-tau rating showed weakened relationships that were only significantly associated with smaller PH volume and ERC thickness. These distinctions are consistent with our hypothesis that quantitative pathology measurements will reduce the ceiling effect observed in semi-quantitative ordinal ratings and provide more granular information about local pathological burden. In the advanced ADNC subgroup, most of the semi-quantitative ratings were very high, while the quantitative measurements showed a large variance in pathological burden, which, according to our results, enhances the ability to predict structural changes beyond coarser stage-based or semi-quantitative measurements. This effect, however, was only seen with tangles; the threads summary measure showed only a few weak but significant associations with antemortem structure. Validation with expert ratings showed poor separation between moderate and severe ratings of thread pathology, suggesting that the threads measurement may be overwhelmed with high amounts of pathology. Of the four pathologies measured quantitatively, the threads measure also showed the lowest accuracy. We also note that, had we used two-tailed tests or a one-sided test for positive associations, a positive association would have been found between the threads measure and AH in the LATE stages 1–3 subgroup. This finding is likely spurious given the small sample size and comparatively weak β coefficients. Future work is needed to understand whether these weak effects are due to limitations of the threads measurement itself or to underlying biology.

The results using quantitative pathology methods for pTDP-43 were less conclusive. Consistent with prior literature [2, 10, 45, 47, 96] we found the strongest associations of pTDP-43 pathology (measured both quantitatively and semi-quantitatively) with AH, PH, and ERC in both our whole cohort and analyses restricted to LATE patients. This pattern was particularly clear in our pointwise regional thickness association maps, where p-tau associations spanned the MTL, but pTDP-43 associations were limited to only the hippocampus and entorhinal cortex. While quantitative p-tau measurements improved upon semi-quantitative ratings in modeling antemortem structural measures, quantifying pTDP-43 did not produce the same advantage. Model comparisons showed only weak evidence that quantitative TDP-43 summary pathology measurements improved model fit for some extrahippocampal regions, and for some regions semi-quantitative measurements were a significantly better model of thickness. We have several hypotheses for this distinction.

First, prior work has shown that in cases with both ADNC and TDP-43 pathology, the vast majority of immunoreactivity is from tau pathology, with only a small amount due to TDP-43 inclusions [63]. Thus, even in cases with LATE-NC, there are likely much fewer pTDP-43 inclusions in a given section than there are p-tau inclusions in an AD case. As a result, our sampling strategy, which did not sample the entire hippocampus, may have missed some of the pTDP-43 pathology that was captured by the more holistic semi-quantitative ratings. Additionally, unlike p-tau, the semi-quantitative pTDP-43 ratings did not show a ceiling effect; variance in both quantitative and semi-quantitative pTDP-43 measures was high, even at severe LATE-NC stages. As a result, the semi-quantitative pTDP-43 ratings may be sufficient to capture the variance associated with thickness.

We also note that all four quantitative measures were associated with some noise, but this may be a larger issue for pTDP-43 analyses because of the imbalance in our cohort. Specifically, only 29 of the 142 cases examined in the pTDP-43 model comparisons had any stage of LATE-NC, and only 14 had LATE-NC stages 2 or 3, where hippocampal pTDP-43 pathology was expected. The noise in the quantitative pTDP-43 measures in the large number of cases with no/low LATE-NC may have drowned out the real signal in the smaller subset of cases that were expected to have any hippocampal pTDP-43 pathology (LATE-NC stage 2/3). The semi-quantitative ratings, however, had little to no noise at low LATE-NC stages, which could improve the strength of this association.

Finally, our quantitative measurements of pTDP-43 in LATE-NC may have also been confounded by the fact that pTDP-43 is associated with neuronal loss and no remnant of the pTDP-43 inclusions after cell death, as opposed to the presence of ghost tangles with p-tau NFTs. Indeed, in FTLD-TDP-43 type C cases, Kawles et al. [52] found that MTL regions showed the highest levels of neurodegeneration but with low densities of TDP-43 inclusions at autopsy; areas without significant neurodegeneration showed higher densities of TDP-43 inclusions. This association might result in lower amounts of pTDP-43 in cases with more severe neuronal loss which may blunt relationships between structure and total pTDP-43 load. Staging schemes, which are a measure of extent, may be less sensitive to this process.

There are also some statistical reasons for this distinction: the model comparisons were influenced by differences in the number of parameters in the models. The semi-quantitative model only had one pTDP-43 variable (semi-quantitative MTL pTDP-43 rating) while the quantitative model had two (neuronal/glial and neuritic pTDP-43 summary measures), and both BIC and AICc penalize models with more parameters. Including only the neuronal/glial pTDP-43 variable in the quantitative model slightly improved the model relative to the semi-quantitative model, but not substantially.

This study builds on findings from our previous work [27, 95]. We have an overlapping but distinct sample from these studies with newer autopsies included, cases with any neuropathological diagnosis of non-AD tauopathies and FTLD excluded, and only ipsilateral MRI and pathology measures examined. Additionally, we now utilize both semi-quantitative and quantitative measurements of pathology to characterize regional pathology burden. The previous studies did not find as widespread associations with tau as the current study; explanations for this distinction include differences in statistical analysis (such as correction for multiple comparisons and the inclusion of other pathologies), sample differences, and improved ability in the present study to measure p-tau burden via quantitative methods. In addition, de Flores and colleagues [27] analyzed patients with intermediate and high ADNC and Wisse et al. [95] analyzed patients who were beta-amyloid negative, whereas here we examined patients across all levels of ADNC. As a result, we likely had larger variation in the amount of p-tau present; if we limited our sample to only the intermediate and high ADNC cases, like de Flores et al., most of those cases would fall in the Braak V/VI analysis, where we found more limited associations of the semi-quantitative MTL p-tau rating with structure.

Both prior works also showed a clear relationship of TDP-43 pathology to AH atrophy, which has been found in other studies as well [2, 33, 77], and is consistent with descriptions of pathology progression in LATE-NC beginning in the anterior MTL [45, 46, 69]. However, in our study, we observed statistically significant associations of pTDP-43 with AH, PH, and ERC, and there was no clear predilection for preferential association of pTDP-43 with AH. Multiple LATE-NC stage 2/3 cases were excluded from the imaging analyses due to failed segmentation, potentially increasing the likelihood of exclusion of cases with more severe AH atrophy. These exclusions could have contributed to the discrepancies with the previously documented associations of TDP-43 with AH neurodegeneration [2, 27, 33, 77, 95].

There are a few limitations of this study. First, we were unable to analyze associations between structure and pathology at the level of individual ROIs. Due to tearing and staining, not every region was annotated in every histology section, so analyses mapping pathology regions to MRI regions would have substantially reduced our sample size. Instead, we averaged ROI-level pathology measures in the hippocampus for summary quantitative measures. We also pursued slightly different methods of annotations for the two pathologies. For p-tau, novices and experts met and annotated cases in a one-day rating event, and all slides from each specimen were available for annotations. For pTDP-43, novices were trained and supervised by experts and subsequently performed annotations over the course of a few months, and sections specific to different diseases were prioritized for training. [Many of the annotators for the p-tau training event had participated in prior similar training events [103] and were familiar with p-tau pathology. pTDP-43 annotations were new to all other than the experts, so we elected to train a smaller number of researchers over a longer training period to annotate pTDP-43 inclusions]. Differences in WildCat performance are thus likely a result of both biological differences and training protocols.

As stated previously, we also excluded some severely atrophied LATE-NC cases due to mis-segmentation by T1-ASHS. The original data-set used to train T1-ASHS was composed of cognitively normal cases and mildly cognitive impaired patients [104]. Thus, ASHS failed to accurately segment some of the more severely atrophied cases with LATE-NC or more severe AD. These segmentation failures reduced the size of the LATE-NC imaging cohort, which could have contributed to the weaker associations seen with pTDP-43, particularly in the LATE-NC stage 2/3 supplementary analysis where the sample size was limited (N = 14). The rate of LATE-NC in this study (22% in the main analyses with any stage of LATE-NC) was lower than reported in other studies. For example, in a large multi-cohort study of over 6000 participants, Nelson et al. (2022) found that 39% of cases had LATE-NC [67]. This discrepancy is likely attributed to a lower age in our study (average age of 75 years) compared to prior studies (88 in [67]), as LATE-NC is more prevalent in participants of older age [68, 70]. Additionally, we used ordinal ratings of MTL pathology to define LATE-NC stages, but this was not a perfect concordance with current staging guidelines [69] due to inconsistencies in the location of the entorhinal cortex rating (at the level of the amygdala or hippocampal body) and the fact that our semi-quantitative ratings examine all pTDP-43 pathology, not just NCIs. This creates some ambiguity in staging, mostly between stages 1 and 2, but is unlikely to have had a major influence on our results.

We view the use of antemortem MRI as a strength of this study but recognize that other imaging modalities could also be used with distinct advantages and disadvantages for each. Ex vivo MRI allows for much higher resolution images and captures MRI and pathology measures at the same time point. However, ex vivo MRI is not widely performed, and its analysis must cope with changes to the brain due to swelling, brain removal, and fixation. Studies leveraging antemortem MRI are easier to generalize to in vivo biomarkers, but disadvantages include variability in quality, protocol, and availability of antemortem MRI data, as well as potentially long-time intervals between MRI and pathology measures, unless imaging is performed in a terminally ill population [14, 15]. We limit our scans to only those within 10 years before death and covary for this antemortem interval, but this approach cannot fully account for differences in disease progression between the time of antemortem MRI and histopathology. A compromise approach is to scan cadavers before autopsy using in vivo MRI protocols, but this approach is only possible in very few select medical centers [41, 42]. We chose to use antemortem imaging to maximize our sample size and increase the feasibility of generalizing our findings to in vivo biomarkers.

This study also has some additional strengths. First, we used postmortem neuropathology and antemortem MRI of the ipsilateral hemisphere. Our large cohort is well-characterized in terms of neuropathology; we exclude any cases with neuropathological diagnoses of FTLD, ALS, or non-AD tauopathies, such that our results are specific to p-tau and pTDP-43 in the context of AD and LATE-NC. Critically, we utilize both quantitative and semi-quantitative measurements of neuropathology, which allows us to reduce the ceiling effect in severe tau pathology populations seen with semi-quantitative ratings. We considered associations significant at p < 0.05 due to strong a priori hypotheses about the negative relationship between pathology and antemortem structural measures, but the majority of these associations would also survive Bonferroni correction for multiple comparisons (p < 8.33e-03, corresponding to 0.05/6 regions, bolded in each table). While reporting uncorrected results can increase the likelihood of type I errors, the fact that most results here would survive correction for multiple comparisons suggests that chance is small. Future work should expand the sample to include more cases with LATE-NC to re-evaluate our pTDP-43 findings, including the model comparisons, in a larger cohort. Additionally, as many of the cases with LATE-NC had pTDP-43 pathology limited to the amygdala, and TDP-43 has been also shown to be associated with amygdala volume [60], future studies should examine the relationship between imaging and pathology measures of the amygdala as well.

In conclusion, we find strong associations between quantitative measurements of hippocampal p-tau tangles with antemortem MRI measures across MTL subfields, compared to a more limited effect of pTDP-43 to the hippocampus and entorhinal cortex. These different patterns of association can be used to advance in vivo imaging markers of tau and TDP-43 pathology. While quantitative and semi-quantitative measures of pTDP-43 performed similarly, there was a clear improvement in modeling thickness using quantitative measurements of p-tau pathology and reduced ceiling effect in severe disease stages. Taken together, the results demonstrate the power of detailed semi-quantitative and quantitative neuropathology and antemortem neuroimaging analyses to identify neurodegeneration patterns of common co-pathologies and motivate continued investigation to improve characterization of these patterns in vivo.

Supplementary Information

Below is the link to the electronic supplementary material.Supplementary Information (DOCX 11290 KB)

Imaging Analyses Data (CSV 115 KB)

Pathology Analyses Data (CSV 137 KB)

Acknowledgements

We gratefully acknowledge the tissue donors and their families.

Author contributions

A.E.D., P.A.Y., D.A.W., S.R.D., L.E.M.W., R.In., and R.D.F., conceptualized the study and developed the methods. A.E.D, R.It., L.M.L., N.S., C.A., E.C., S.R., M.D., M.T.D., Y.L., A.I., L.P.S., P.S., M.L., A.B., J.Z., B.E.S., R.W., B.K., M.H.S., J.L.R., T.S., D.T.O., S.A., S.A.P., J.A.D, S.R.D., D.J.I., E.B.L., and P.A.Y. performed research to acquire and process the data. A.E.D., P.A.Y., D.A.W, S.R.D., P.K., S.H., C.B., and S.X.X., analyzed the data. A.E.D., P.A.Y., S.R.D., D.A.W., L.E.M.W., R.In., and R.D.F. interpreted the data and discussed the results. A.E.D. drafted the initial manuscript. All authors revised the manuscript and provided critical feedback. P.A.Y. was the principal designer and coordinator of the study and supervised collection, analysis, and interpretation of the study data.

Funding

R01 AG069474, RF1 AG056014, P30 AG072979, P01 AG066597, U19 AG062418, R01 AG075276, R01 AG070592, R01-NS109260. L.E.M.W. was supported by MultiPark—A Strategic Research Area at Lund University. L.P.S. was supported by T32 AG076411. C.B. was supported by R25-NS065745 and support from the Alzheimer’s Association and Fred A. and Barbara M. Erb Family Foundation (AACSF-23-1152241). S.H was supported by AARF-21- 48972.

Availability of data and material

Pathology and imaging measurements have been uploaded as supplementary material and to Dryad (https://doi.org/10.5061/dryad.95x69p8tq). Images will be uploaded to the same depository pending anonymization.

Declarations

Conflict of interest

D.A.W has served as a paid consultant to Eli Lilly, GE Healthcare, and Qynapse. He serves on a DSMB for Functional Neuromodulation and GSK. He is a site investigator for a clinical trial sponsored by Biogen. D.J.I. is member of science advisory board of Denali Therapeutics. S.R.D. received consultation fees from Rancho Biosciences and Nia Therapeutics. The other authors have nothing to disclose.

Ethics approval and consent to participate

Where possible, pre-consent during life and, in all cases, next-of-kin consent at death was given. Imaging studies were approved by the Institutional Review Board at the University of Pennsylvania and were in accordance with the Declaration of Helsinki.

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
==== Refs
References

1. Barnes J Ridgway GR Bartlett J Henley SMD Lehmann M Hobbs N Head size, age and gender adjustment in MRI studies: a necessary nuisance? NeuroImage 2010 53 1244 1255 10.1016/j.neuroimage.2010.06.025 20600995
Barnes J, Ridgway GR, Bartlett J, Henley SMD, Lehmann M, Hobbs N et al (2010) Head size, age and gender adjustment in MRI studies: a necessary nuisance? NeuroImage 53:1244–1255. 10.1016/j.neuroimage.2010.06.02520600995 10.1016/j.neuroimage.2010.06.025
2. Bejanin A Murray ME Martin P Botha H Tosakulwong N Schwarz CG Antemortem volume loss mirrors TDP-43 staging in older adults with non-frontotemporal lobar degeneration Brain J Neurol 2019 142 3621 3635 10.1093/brain/awz277
Bejanin A, Murray ME, Martin P, Botha H, Tosakulwong N, Schwarz CG et al (2019) Antemortem volume loss mirrors TDP-43 staging in older adults with non-frontotemporal lobar degeneration. Brain J Neurol 142:3621–3635. 10.1093/brain/awz27710.1093/brain/awz277
3. Berron D Vogel JW Insel PS Pereira JB Xie L Wisse LEM Early stages of tau pathology and its associations with functional connectivity, atrophy and memory Brain 2021 144 2771 2783 10.1093/brain/awab114 33725124
Berron D, Vogel JW, Insel PS, Pereira JB, Xie L, Wisse LEM et al (2021) Early stages of tau pathology and its associations with functional connectivity, atrophy and memory. Brain 144:2771–2783. 10.1093/brain/awab11433725124 10.1093/brain/awab114
4. Botha H Mantyh WG Murray ME Knopman DS Przybelski SA Wiste HJ FDG-PET in tau-negative amnestic dementia resembles that of autopsy-proven hippocampal sclerosis Brain 2018 141 1201 1217 10.1093/brain/awy049 29538658
Botha H, Mantyh WG, Murray ME, Knopman DS, Przybelski SA, Wiste HJ et al (2018) FDG-PET in tau-negative amnestic dementia resembles that of autopsy-proven hippocampal sclerosis. Brain 141:1201–1217. 10.1093/brain/awy04929538658 10.1093/brain/awy049
5. Braak H Braak E Neuropathological stageing of Alzheimer-related changes Acta Neuropathol (Berl) 1991 82 239 259 10.1007/BF00308809 1759558
Braak H, Braak E (1991) Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol (Berl) 82:239–259. 10.1007/BF003088091759558 10.1007/BF00308809
6. Braak H Braak E Staging of alzheimer’s disease-related neurofibrillary changes Neurobiol Aging 1995 16 271 278 10.1016/0197-4580(95)00021-6 7566337
Braak H, Braak E (1995) Staging of alzheimer’s disease-related neurofibrillary changes. Neurobiol Aging 16:271–278. 10.1016/0197-4580(95)00021-67566337 10.1016/0197-4580(95)00021-6
7. Brettschneider J Del Tredici K Irwin DJ Grossman M Robinson JL Toledo JB Sequential distribution of pTDP-43 pathology in behavioral variant frontotemporal dementia (bvFTD) Acta Neuropathol (Berl) 2014 127 423 439 10.1007/s00401-013-1238-y 24407427
Brettschneider J, Del Tredici K, Irwin DJ, Grossman M, Robinson JL, Toledo JB et al (2014) Sequential distribution of pTDP-43 pathology in behavioral variant frontotemporal dementia (bvFTD). Acta Neuropathol (Berl) 127:423–439. 10.1007/s00401-013-1238-y24407427 10.1007/s00401-013-1238-y
8. Brettschneider J Del Tredici K Toledo JB Robinson JL Irwin DJ Grossman M Stages of pTDP-43 pathology in amyotrophic lateral sclerosis Ann Neurol 2013 74 20 38 10.1002/ana.23937 23686809
Brettschneider J, Del Tredici K, Toledo JB, Robinson JL, Irwin DJ, Grossman M et al (2013) Stages of pTDP-43 pathology in amyotrophic lateral sclerosis. Ann Neurol 74:20–38. 10.1002/ana.2393723686809 10.1002/ana.23937
9. Buciuc M Botha H Murray ME Schwarz CG Senjem ML Jones DT Utility of FDG-PET in diagnosis of Alzheimer-related TDP-43 proteinopathy Neurology 2020 95 e23 e34 10.1212/WNL.0000000000009722 32518145
Buciuc M, Botha H, Murray ME, Schwarz CG, Senjem ML, Jones DT et al (2020) Utility of FDG-PET in diagnosis of Alzheimer-related TDP-43 proteinopathy. Neurology 95:e23–e34. 10.1212/WNL.000000000000972232518145 10.1212/WNL.0000000000009722
10. Buciuc M Wennberg AM Weigand SD Murray ME Senjem ML Spychalla AJ Effect modifiers of TDP-43-associated hippocampal atrophy rates in patients with Alzheimer’s disease neuropathological changes J Alzheimers Dis 2020 73 1511 1523 10.3233/JAD-191040 31929165
Buciuc M, Wennberg AM, Weigand SD, Murray ME, Senjem ML, Spychalla AJ et al (2020) Effect modifiers of TDP-43-associated hippocampal atrophy rates in patients with Alzheimer’s disease neuropathological changes. J Alzheimers Dis 73:1511–1523. 10.3233/JAD-19104031929165 10.3233/JAD-191040
11. Cairns NJ Bigio EH Mackenzie IRA Neumann M Lee VM-Y Hatanpaa KJ Neuropathologic diagnostic and nosologic criteria for frontotemporal lobar degeneration: consensus of the Consortium for Frontotemporal Lobar Degeneration Acta Neuropathol (Berl) 2007 114 5 22 10.1007/s00401-007-0237-2 17579875
Cairns NJ, Bigio EH, Mackenzie IRA, Neumann M, Lee VM-Y, Hatanpaa KJ et al (2007) Neuropathologic diagnostic and nosologic criteria for frontotemporal lobar degeneration: consensus of the Consortium for Frontotemporal Lobar Degeneration. Acta Neuropathol (Berl) 114:5–22. 10.1007/s00401-007-0237-217579875 10.1007/s00401-007-0237-2
12. Chatterjee M, Özdemir S, Fritz C, Möbius W, Kleineidam L, Mandelkow E, et al. (2023) Plasma extracellular vesicle Tau isoform ratios and TDP-43 inform about molecular pathology in Frontotemporal Dementia and Amyotrophic Lateral Sclerosis
13. Chauveau L Kuhn E Palix C Felisatti F Ourry V de La Sayette V Medial temporal lobe subregional atrophy in aging and Alzheimer’s disease: a longitudinal study Front Aging Neurosci 2021 13 750154 10.3389/fnagi.2021.750154 34720998
Chauveau L, Kuhn E, Palix C, Felisatti F, Ourry V, de La Sayette V et al (2021) Medial temporal lobe subregional atrophy in aging and Alzheimer’s disease: a longitudinal study. Front Aging Neurosci 13:75015434720998 10.3389/fnagi.2021.750154
14. Clark CM Pontecorvo MJ Beach TG Bedell BJ Coleman RE Doraiswamy PM Cerebral PET with florbetapir compared with neuropathology at autopsy for detection of neuritic amyloid-β plaques: a prospective cohort study Lancet Neurol 2012 11 669 678 10.1016/S1474-4422(12)70142-4 22749065
Clark CM, Pontecorvo MJ, Beach TG, Bedell BJ, Coleman RE, Doraiswamy PM et al (2012) Cerebral PET with florbetapir compared with neuropathology at autopsy for detection of neuritic amyloid-β plaques: a prospective cohort study. Lancet Neurol 11:669–678. 10.1016/S1474-4422(12)70142-422749065 10.1016/S1474-4422(12)70142-4
15. Clark CM Schneider JA Bedell BJ Beach TG Bilker WB Mintun MA Use of florbetapir-PET for imaging β-amyloid pathology JAMA 2011 305 275 283 10.1001/jama.2010.2008 21245183
Clark CM, Schneider JA, Bedell BJ, Beach TG, Bilker WB, Mintun MA et al (2011) Use of florbetapir-PET for imaging β-amyloid pathology. JAMA 305:275–283. 10.1001/jama.2010.200821245183 10.1001/jama.2010.2008
16. Coughlin D Xie SX Liang M Williams A Peterson C Weintraub D Cognitive and pathological influences of tau pathology in lewy body disorders Ann Neurol 2019 85 259 271 10.1002/ana.25392 30549331
Coughlin D, Xie SX, Liang M, Williams A, Peterson C, Weintraub D et al (2019) Cognitive and pathological influences of tau pathology in lewy body disorders. Ann Neurol 85:259–271. 10.1002/ana.2539230549331 10.1002/ana.25392
17. Cummings JL Henchcliffe C Schaier S Simuni T Waxman A Kemp P The role of dopaminergic imaging in patients with symptoms of dopaminergic system neurodegeneration Brain J Neurol 2011 134 3146 3166 10.1093/brain/awr177
Cummings JL, Henchcliffe C, Schaier S, Simuni T, Waxman A, Kemp P (2011) The role of dopaminergic imaging in patients with symptoms of dopaminergic system neurodegeneration. Brain J Neurol 134:3146–3166. 10.1093/brain/awr17710.1093/brain/awr177
18. Dallaire-Théroux C Beheshti I Potvin O Dieumegarde L Saikali S Duchesne S Braak neurofibrillary tangle staging prediction from in vivo MRI metrics Alzheimers Dement Diagn Assess Dis Monit 2019 11 599 609 10.1016/j.dadm.2019.07.001
Dallaire-Théroux C, Beheshti I, Potvin O, Dieumegarde L, Saikali S, Duchesne S (2019) Braak neurofibrillary tangle staging prediction from in vivo MRI metrics. Alzheimers Dement Diagn Assess Dis Monit 11:599–609. 10.1016/j.dadm.2019.07.00110.1016/j.dadm.2019.07.001
19. Dallaire-Théroux C Callahan BL Potvin O Saikali S Duchesne S Radiological-pathological correlation in Alzheimer’s disease: systematic review of antemortem magnetic resonance imaging findings J Alzheimers Dis JAD 2017 57 575 601 10.3233/JAD-161028 28282807
Dallaire-Théroux C, Callahan BL, Potvin O, Saikali S, Duchesne S (2017) Radiological-pathological correlation in Alzheimer’s disease: systematic review of antemortem magnetic resonance imaging findings. J Alzheimers Dis JAD 57:575–601. 10.3233/JAD-16102828282807 10.3233/JAD-161028
20. Das SR Xie L Wisse LEM Vergnet N Ittyerah R Cui S In-vivo measures of tau burden are associated with atrophy in early braak stage medial temporal lobe regions in amyloid negative individuals Alzheimers Dement J Alzheimers Assoc 2019 15 1286 1295 10.1016/j.jalz.2019.05.009
Das SR, Xie L, Wisse LEM, Vergnet N, Ittyerah R, Cui S et al (2019) In-vivo measures of tau burden are associated with atrophy in early braak stage medial temporal lobe regions in amyloid negative individuals. Alzheimers Dement J Alzheimers Assoc 15:1286–1295. 10.1016/j.jalz.2019.05.00910.1016/j.jalz.2019.05.009
21. Duong MT Das SR Lyu X Xie L Richardson H Xie SX Dissociation of tau pathology and neuronal hypometabolism within the ATN framework of Alzheimer’s disease Nat Commun 2022 13 1495 10.1038/s41467-022-28941-1 35314672
Duong MT, Das SR, Lyu X, Xie L, Richardson H, Xie SX et al (2022) Dissociation of tau pathology and neuronal hypometabolism within the ATN framework of Alzheimer’s disease. Nat Commun 13:1495. 10.1038/s41467-022-28941-135314672 10.1038/s41467-022-28941-1
22. Durand T, Mordan T, Thome N, Cord M (2017) WILDCAT: weakly supervised learning of deep ConvNets for image classification, pointwise localization and segmentation. In: 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). pp 5957–5966
23. van Dyck CH Swanson CJ Aisen P Bateman RJ Chen C Gee M Lecanemab in early Alzheimer’s disease N Engl J Med 2023 388 9 21 10.1056/NEJMoa2212948 36449413
van Dyck CH, Swanson CJ, Aisen P, Bateman RJ, Chen C, Gee M et al (2023) Lecanemab in early Alzheimer’s disease. N Engl J Med 388:9–21. 10.1056/NEJMoa221294836449413 10.1056/NEJMoa2212948
24. Ferman TJ Aoki N Crook JE Murray ME Graff-Radford NR van Gerpen JA The limbic and neocortical contribution of α-synuclein, tau, and amyloid β to disease duration in dementia with Lewy bodies Alzheimers Dement J Alzheimers Assoc 2018 14 330 339 10.1016/j.jalz.2017.09.014
Ferman TJ, Aoki N, Crook JE, Murray ME, Graff-Radford NR, van Gerpen JA et al (2018) The limbic and neocortical contribution of α-synuclein, tau, and amyloid β to disease duration in dementia with Lewy bodies. Alzheimers Dement J Alzheimers Assoc 14:330–339. 10.1016/j.jalz.2017.09.01410.1016/j.jalz.2017.09.014
25. Ferrer I López-González I Carmona M Arregui L Dalfó E Torrejón-Escribano B Glial and neuronal tau pathology in tauopathies: characterization of disease-specific phenotypes and tau pathology progression J Neuropathol Exp Neurol 2014 73 81 97 10.1097/NEN.0000000000000030 24335532
Ferrer I, López-González I, Carmona M, Arregui L, Dalfó E, Torrejón-Escribano B et al (2014) Glial and neuronal tau pathology in tauopathies: characterization of disease-specific phenotypes and tau pathology progression. J Neuropathol Exp Neurol 73:81–97. 10.1097/NEN.000000000000003024335532 10.1097/NEN.0000000000000030
26. Ferrer I Santpere G van Leeuwen FW Argyrophilic grain disease Brain J Neurol 2008 131 1416 1432 10.1093/brain/awm305
Ferrer I, Santpere G, van Leeuwen FW (2008) Argyrophilic grain disease. Brain J Neurol 131:1416–1432. 10.1093/brain/awm30510.1093/brain/awm305
27. de Flores R Wisse LEM Das SR Xie L McMillan CT Trojanowski JQ Contribution of mixed pathology to medial temporal lobe atrophy in Alzheimer’s disease Alzheimers Dement 2020 16 843 852 10.1002/alz.12079 32323446
de Flores R, Wisse LEM, Das SR, Xie L, McMillan CT, Trojanowski JQ et al (2020) Contribution of mixed pathology to medial temporal lobe atrophy in Alzheimer’s disease. Alzheimers Dement 16:843–852. 10.1002/alz.1207932323446 10.1002/alz.12079
28. Freedman D Lane D A nonstochastic interpretation of reported significance levels J Bus Econ Stat 1983 1 292 298 10.2307/1391660
Freedman D, Lane D (1983) A nonstochastic interpretation of reported significance levels. J Bus Econ Stat 1:292–298. 10.2307/139166010.2307/1391660
29. von Gunten A Kövari E Bussière T Rivara C-B Gold G Bouras C Hof PR Cognitive impact of neuronal pathology in the entorhinal cortex and CA1 field in Alzheimer’s disease Neurobiol Aging 2006 27 270 277 10.1016/j.neurobiolaging.2005.02.008 16399212
von Gunten A, Kövari E, Bussière T, Rivara C-B, Gold G, Bouras C et al (2006) Cognitive impact of neuronal pathology in the entorhinal cortex and CA1 field in Alzheimer’s disease. Neurobiol Aging 27:270–277. 10.1016/j.neurobiolaging.2005.02.00816399212 10.1016/j.neurobiolaging.2005.02.008
30. Hall S Orrù CD Serrano GE Galasko D Hughson AG Groveman BR Performance of αSynuclein RT-QuIC in relation to neuropathological staging of Lewy body disease Acta Neuropathol Commun 2022 10 90 10.1186/s40478-022-01388-7 35733234
Hall S, Orrù CD, Serrano GE, Galasko D, Hughson AG, Groveman BR et al (2022) Performance of αSynuclein RT-QuIC in relation to neuropathological staging of Lewy body disease. Acta Neuropathol Commun 10:90. 10.1186/s40478-022-01388-735733234 10.1186/s40478-022-01388-7
31. Hanko V Apple AC Alpert KI Warren KN Schneider JA Arfanakis K In vivo hippocampal subfield shape related to TDP-43, amyloid beta, and tau pathologies Neurobiol Aging 2019 74 171 181 10.1016/j.neurobiolaging.2018.10.013 30453234
Hanko V, Apple AC, Alpert KI, Warren KN, Schneider JA, Arfanakis K et al (2019) In vivo hippocampal subfield shape related to TDP-43, amyloid beta, and tau pathologies. Neurobiol Aging 74:171–181. 10.1016/j.neurobiolaging.2018.10.01330453234 10.1016/j.neurobiolaging.2018.10.013
32. Hayasaka S Nichols TE Combining voxel intensity and cluster extent with permutation test framework Neuroimage 2004 23 54 63 10.1016/j.neuroimage.2004.04.035 15325352
Hayasaka S, Nichols TE (2004) Combining voxel intensity and cluster extent with permutation test framework. Neuroimage 23:54–63. 10.1016/j.neuroimage.2004.04.03515325352 10.1016/j.neuroimage.2004.04.035
33. Heywood A Stocks J Schneider JA Arfanakis K Bennett DA Beg MF The unique effect of TDP-43 on hippocampal subfield morphometry and cognition NeuroImage Clin 2022 35 103125 10.1016/j.nicl.2022.103125 36002965
Heywood A, Stocks J, Schneider JA, Arfanakis K, Bennett DA, Beg MF et al (2022) The unique effect of TDP-43 on hippocampal subfield morphometry and cognition. NeuroImage Clin 35:103125. 10.1016/j.nicl.2022.10312536002965 10.1016/j.nicl.2022.103125
34. Hrybouski S Das SR Xie L Wisse LEM Kelley M Lane J Aging and Alzheimer’s disease have dissociable effects on local and regional medial temporal lobe connectivity Brain Commun 2023 5 fcad245 10.1093/braincomms/fcad245 37767219
Hrybouski S, Das SR, Xie L, Wisse LEM, Kelley M, Lane J et al (2023) Aging and Alzheimer’s disease have dissociable effects on local and regional medial temporal lobe connectivity. Brain Commun 5:fcad245. 10.1093/braincomms/fcad24537767219 10.1093/braincomms/fcad245
35. Huntenburg JM Steele CJ Bazin P-L Nighres: processing tools for high-resolution neuroimaging GigaScience 2018 7 giy082 10.1093/gigascience/giy082 29982501
Huntenburg JM, Steele CJ, Bazin P-L (2018) Nighres: processing tools for high-resolution neuroimaging. GigaScience 7:giy082. 10.1093/gigascience/giy08229982501 10.1093/gigascience/giy082
36. Iglesias JE Augustinack JC Nguyen K Player CM Player A Wright M A computational atlas of the hippocampal formation using ex vivo, ultra-high resolution MRI: application to adaptive segmentation of in vivo MRI Neuroimage 2015 115 117 137 10.1016/j.neuroimage.2015.04.042 25936807
Iglesias JE, Augustinack JC, Nguyen K, Player CM, Player A, Wright M et al (2015) A computational atlas of the hippocampal formation using ex vivo, ultra-high resolution MRI: application to adaptive segmentation of in vivo MRI. Neuroimage 115:117–137. 10.1016/j.neuroimage.2015.04.04225936807 10.1016/j.neuroimage.2015.04.042
37. Irwin DJ Byrne MD McMillan CT Cooper F Arnold SE Lee EB Semi-automated digital image analysis of pick’s disease and TDP-43 proteinopathy J Histochem Cytochem 2016 64 54 66 10.1369/0022155415614303 26538548
Irwin DJ, Byrne MD, McMillan CT, Cooper F, Arnold SE, Lee EB et al (2016) Semi-automated digital image analysis of pick’s disease and TDP-43 proteinopathy. J Histochem Cytochem 64:54–66. 10.1369/002215541561430326538548 10.1369/0022155415614303
38. Irwin DJ McMillan CT Brettschneider J Libon DJ Powers J Rascovsky K Cognitive decline and reduced survival in C9orf72 expansion frontotemporal degeneration and amyotrophic lateral sclerosis J Neurol Neurosurg Psychiatry 2013 84 163 169 10.1136/jnnp-2012-303507 23117491
Irwin DJ, McMillan CT, Brettschneider J, Libon DJ, Powers J, Rascovsky K et al (2013) Cognitive decline and reduced survival in C9orf72 expansion frontotemporal degeneration and amyotrophic lateral sclerosis. J Neurol Neurosurg Psychiatry 84:163–169. 10.1136/jnnp-2012-30350723117491 10.1136/jnnp-2012-303507
39. James BD Wilson RS Boyle PA Trojanowski JQ Bennett DA Schneider JA TDP-43 stage, mixed pathologies, and clinical Alzheimer’s-type dementia Brain J Neurol 2016 139 2983 2993 10.1093/brain/aww224
James BD, Wilson RS, Boyle PA, Trojanowski JQ, Bennett DA, Schneider JA (2016) TDP-43 stage, mixed pathologies, and clinical Alzheimer’s-type dementia. Brain J Neurol 139:2983–2993. 10.1093/brain/aww22410.1093/brain/aww224
40. Jarholm JA Bjørnerud A Dalaker TO Akhavi MS Kirsebom BE Pålhaugen L Medial temporal lobe atrophy in predementia Alzheimer’s disease: a longitudinal multi-site study comparing staging and A/T/N in a clinical research cohort1 J Alzheimers Dis 2023 94 259 279 10.3233/JAD-221274 37248900
Jarholm JA, Bjørnerud A, Dalaker TO, Akhavi MS, Kirsebom BE, Pålhaugen L et al (2023) Medial temporal lobe atrophy in predementia Alzheimer’s disease: a longitudinal multi-site study comparing staging and A/T/N in a clinical research cohort1. J Alzheimers Dis 94:259–279. 10.3233/JAD-22127437248900 10.3233/JAD-221274
41. Jonkman LE Reijner N Frigerio I Bouwman MMA Lin C-P Rozemuller AJM Parietal-temporal structural network reorganization is associated with pathological load and axonal damage in Alzheimer’s disease: a post-mortem MRI and pathology study Alzheimers Dement 2023 19 e079516 10.1002/alz.079516
Jonkman LE, Reijner N, Frigerio I, Bouwman MMA, Lin C-P, Rozemuller AJM et al (2023) Parietal-temporal structural network reorganization is associated with pathological load and axonal damage in Alzheimer’s disease: a post-mortem MRI and pathology study. Alzheimers Dement 19:e079516. 10.1002/alz.07951610.1002/alz.079516
42. Jonkman LE Steenwijk MD Boesen N Rozemuller AJM Barkhof F Geurts JJG Relationship between β-amyloid and structural network topology in decedents without dementia Neurology 2020 95 e532 e544 10.1212/WNL.0000000000009910 32661099
Jonkman LE, Steenwijk MD, Boesen N, Rozemuller AJM, Barkhof F, Geurts JJG et al (2020) Relationship between β-amyloid and structural network topology in decedents without dementia. Neurology 95:e532–e544. 10.1212/WNL.000000000000991032661099 10.1212/WNL.0000000000009910
43. Josephs KA Dickson DW Tosakulwong N Weigand SD Murray ME Petrucelli L Rates of hippocampal atrophy and presence of post-mortem TDP-43 in patients with Alzheimer’s disease: a longitudinal retrospective study Lancet Neurol 2017 16 917 924 10.1016/S1474-4422(17)30284-3 28919059
Josephs KA, Dickson DW, Tosakulwong N, Weigand SD, Murray ME, Petrucelli L et al (2017) Rates of hippocampal atrophy and presence of post-mortem TDP-43 in patients with Alzheimer’s disease: a longitudinal retrospective study. Lancet Neurol 16:917–924. 10.1016/S1474-4422(17)30284-328919059 10.1016/S1474-4422(17)30284-3
44. Josephs KA Murray ME Tosakulwong N Weigand SD Serie AM Perkerson RB Pathological, imaging and genetic characteristics support the existence of distinct TDP-43 types in non-FTLD brains Acta Neuropathol (Berl) 2019 137 227 238 10.1007/s00401-018-1951-7 30604226
Josephs KA, Murray ME, Tosakulwong N, Weigand SD, Serie AM, Perkerson RB et al (2019) Pathological, imaging and genetic characteristics support the existence of distinct TDP-43 types in non-FTLD brains. Acta Neuropathol (Berl) 137:227–238. 10.1007/s00401-018-1951-730604226 10.1007/s00401-018-1951-7
45. Josephs KA Murray ME Whitwell JL Parisi JE Petrucelli L Jack CR Staging TDP-43 pathology in Alzheimer’s disease Acta Neuropathol (Berl) 2014 127 441 450 10.1007/s00401-013-1211-9 24240737
Josephs KA, Murray ME, Whitwell JL, Parisi JE, Petrucelli L, Jack CR et al (2014) Staging TDP-43 pathology in Alzheimer’s disease. Acta Neuropathol (Berl) 127:441–450. 10.1007/s00401-013-1211-924240737 10.1007/s00401-013-1211-9
46. Josephs KA Murray ME Whitwell JL Tosakulwong N Weigand SD Petrucelli L Updated TDP-43 in Alzheimer’s disease staging scheme Acta Neuropathol (Berl) 2016 131 571 585 10.1007/s00401-016-1537-1 26810071
Josephs KA, Murray ME, Whitwell JL, Tosakulwong N, Weigand SD, Petrucelli L et al (2016) Updated TDP-43 in Alzheimer’s disease staging scheme. Acta Neuropathol (Berl) 131:571–585. 10.1007/s00401-016-1537-126810071 10.1007/s00401-016-1537-1
47. Josephs KA Whitwell JL Weigand SD Murray ME Tosakulwong N Liesinger AM TDP-43 is a key player in the clinical features associated with Alzheimer’s disease Acta Neuropathol (Berl) 2014 127 811 824 10.1007/s00401-014-1269-z 24659241
Josephs KA, Whitwell JL, Weigand SD, Murray ME, Tosakulwong N, Liesinger AM et al (2014) TDP-43 is a key player in the clinical features associated with Alzheimer’s disease. Acta Neuropathol (Berl) 127:811–824. 10.1007/s00401-014-1269-z24659241 10.1007/s00401-014-1269-z
48. Jung Y Jordan LG Lowe VJ Kantarci K Parisi JE Dickson DW Clinicopathological and 123I-FP-CIT SPECT correlations in patients with dementia Ann Clin Transl Neurol 2018 5 376 381 10.1002/acn3.521 29560382
Jung Y, Jordan LG, Lowe VJ, Kantarci K, Parisi JE, Dickson DW et al (2018) Clinicopathological and 123I-FP-CIT SPECT correlations in patients with dementia. Ann Clin Transl Neurol 5:376–381. 10.1002/acn3.52129560382 10.1002/acn3.521
49. Kantarci K Lowe VJ Boeve BF Weigand SD Senjem ML Przybelski SA Multimodality imaging characteristics of dementia with Lewy bodies Neurobiol Aging 2012 33 2091 2105 10.1016/j.neurobiolaging.2011.09.024 22018896
Kantarci K, Lowe VJ, Boeve BF, Weigand SD, Senjem ML, Przybelski SA et al (2012) Multimodality imaging characteristics of dementia with Lewy bodies. Neurobiol Aging 33:2091–2105. 10.1016/j.neurobiolaging.2011.09.02422018896 10.1016/j.neurobiolaging.2011.09.024
50. Kapasi A Yu L Boyle PA Barnes LL Bennett DA Schneider JA Limbic-predominant age-related TDP-43 encephalopathy, ADNC pathology, and cognitive decline in aging Neurology 2020 95 e1951 e1962 10.1212/WNL.0000000000010454 32753441
Kapasi A, Yu L, Boyle PA, Barnes LL, Bennett DA, Schneider JA (2020) Limbic-predominant age-related TDP-43 encephalopathy, ADNC pathology, and cognitive decline in aging. Neurology 95:e1951–e1962. 10.1212/WNL.000000000001045432753441 10.1212/WNL.0000000000010454
51. Karypis G Kumar V A fast and high quality multilevel scheme for partitioning irregular graphs SIAM J Sci Comput 1998 20 359 392 10.1137/S1064827595287997
Karypis G, Kumar V (1998) A fast and high quality multilevel scheme for partitioning irregular graphs. SIAM J Sci Comput 20:359–392. 10.1137/S106482759528799710.1137/S1064827595287997
52. Kawles A Nishihira Y Feldman A Gill N Minogue G Keszycki R Cortical and subcortical pathological burden and neuronal loss in an autopsy series of FTLD-TDP-type C Brain 2021 145 1069 1078 10.1093/brain/awab368
Kawles A, Nishihira Y, Feldman A, Gill N, Minogue G, Keszycki R et al (2021) Cortical and subcortical pathological burden and neuronal loss in an autopsy series of FTLD-TDP-type C. Brain 145:1069–1078. 10.1093/brain/awab36810.1093/brain/awab368
53. Kotrotsou A Schneider JA Bennett DA Leurgans SE Dawe RJ Boyle PA Neuropathologic correlates of regional brain volumes in a community cohort of older adults Neurobiol Aging 2015 36 2798 2805 10.1016/j.neurobiolaging.2015.06.025 26195068
Kotrotsou A, Schneider JA, Bennett DA, Leurgans SE, Dawe RJ, Boyle PA et al (2015) Neuropathologic correlates of regional brain volumes in a community cohort of older adults. Neurobiol Aging 36:2798–2805. 10.1016/j.neurobiolaging.2015.06.02526195068 10.1016/j.neurobiolaging.2015.06.025
54. Kovacs GG Ferrer I Grinberg LT Alafuzoff I Attems J Budka H Aging-related tau astrogliopathy (ARTAG): harmonized evaluation strategy Acta Neuropathol (Berl) 2016 131 87 102 10.1007/s00401-015-1509-x 26659578
Kovacs GG, Ferrer I, Grinberg LT, Alafuzoff I, Attems J, Budka H et al (2016) Aging-related tau astrogliopathy (ARTAG): harmonized evaluation strategy. Acta Neuropathol (Berl) 131:87–102. 10.1007/s00401-015-1509-x26659578 10.1007/s00401-015-1509-x
55. Kovacs GG Milenkovic I Wöhrer A Höftberger R Gelpi E Haberler C Non-Alzheimer neurodegenerative pathologies and their combinations are more frequent than commonly believed in the elderly brain: a community-based autopsy series Acta Neuropathol (Berl) 2013 126 365 384 10.1007/s00401-013-1157-y 23900711
Kovacs GG, Milenkovic I, Wöhrer A, Höftberger R, Gelpi E, Haberler C et al (2013) Non-Alzheimer neurodegenerative pathologies and their combinations are more frequent than commonly believed in the elderly brain: a community-based autopsy series. Acta Neuropathol (Berl) 126:365–384. 10.1007/s00401-013-1157-y23900711 10.1007/s00401-013-1157-y
56. Lee CS Latimer CS Henriksen JC Blazes M Larson EB Application of deep learning to understand resilience to Alzheimer’s disease pathology Brain Pathol 2021 31 e12974 10.1111/bpa.12974 34009663
Lee CS, Latimer CS, Henriksen JC, Blazes M, Larson EB et al (2021) Application of deep learning to understand resilience to Alzheimer’s disease pathology. Brain Pathol 31:e12974. 10.1111/bpa.1297434009663 10.1111/bpa.12974
57. Lee EB Porta S Michael Baer G Xu Y Suh E Expansion of the classification of FTLD-TDP: distinct pathology associated with rapidly progressive frontotemporal degeneration Acta Neuropathol (Berl) 2017 134 65 78 10.1007/s00401-017-1679-9 28130640
Lee EB, Porta S, Michael Baer G, Xu Y, Suh E et al (2017) Expansion of the classification of FTLD-TDP: distinct pathology associated with rapidly progressive frontotemporal degeneration. Acta Neuropathol (Berl) 134:65–78. 10.1007/s00401-017-1679-928130640 10.1007/s00401-017-1679-9
58. Lim SM Katsifis A Villemagne VL Best R Jones G Saling M The 18F-FDG PET cingulate island sign and comparison to 123I-beta-CIT SPECT for diagnosis of dementia with Lewy bodies J Nucl Med Off Publ Soc Nucl Med 2009 50 1638 1645 10.2967/jnumed.109.065870
Lim SM, Katsifis A, Villemagne VL, Best R, Jones G, Saling M et al (2009) The 18F-FDG PET cingulate island sign and comparison to 123I-beta-CIT SPECT for diagnosis of dementia with Lewy bodies. J Nucl Med Off Publ Soc Nucl Med 50:1638–1645. 10.2967/jnumed.109.06587010.2967/jnumed.109.065870
59. Lyu X Duong MT Xie L de Flores R Richardson H Hwang G For the Alzheimer’s Disease Neuroimaging Initiative Tau-neurodegeneration mismatch reveals vulnerability and resilience to comorbidities in Alzheimer’s continuum Alzheimers Dement 2024 20 1586 1600 10.1002/alz.13559 38050662
Lyu X, Duong MT, Xie L, de Flores R, Richardson H, Hwang G et al (2024) Tau-neurodegeneration mismatch reveals vulnerability and resilience to comorbidities in Alzheimer’s continuum. Alzheimers Dement 20:1586–1600. 10.1002/alz.1355938050662 10.1002/alz.13559
60. Makkinejad N Schneider JA Yu J Leurgans SE Kotrotsou A Evia AM Associations of amygdala volume and shape with transactive response DNA-binding protein 43 (TDP-43) pathology in a community cohort of older adults Neurobiol Aging 2019 77 104 111 10.1016/j.neurobiolaging.2019.01.022 30784812
Makkinejad N, Schneider JA, Yu J, Leurgans SE, Kotrotsou A, Evia AM et al (2019) Associations of amygdala volume and shape with transactive response DNA-binding protein 43 (TDP-43) pathology in a community cohort of older adults. Neurobiol Aging 77:104–111. 10.1016/j.neurobiolaging.2019.01.02230784812 10.1016/j.neurobiolaging.2019.01.022
61. Marx GA Koenigsberg DG McKenzie AT Kauffman J Hanson RW Whitney K The PART working group Artificial intelligence-derived neurofibrillary tangle burden is associated with antemortem cognitive impairment Acta Neuropathol Commun. 2022 10 157 10.1186/s40478-022-01457-x 36316708
Marx GA, Koenigsberg DG, McKenzie AT, Kauffman J, Hanson RW, Whitney K et al (2022) Artificial intelligence-derived neurofibrillary tangle burden is associated with antemortem cognitive impairment. Acta Neuropathol Commun. 10:157. 10.1186/s40478-022-01457-x36316708 10.1186/s40478-022-01457-x
62. McKeith IG Boeve BF Dickson DW Halliday G Taylor J-P Weintraub D Diagnosis and management of dementia with Lewy bodies: fourth consensus report of the DLB Consortium Neurology 2017 89 88 100 10.1212/WNL.0000000000004058 28592453
McKeith IG, Boeve BF, Dickson DW, Halliday G, Taylor J-P, Weintraub D et al (2017) Diagnosis and management of dementia with Lewy bodies: fourth consensus report of the DLB Consortium. Neurology 89:88–100. 10.1212/WNL.000000000000405828592453 10.1212/WNL.0000000000004058
63. Minogue G Kawles A Zouridakis A Keszycki R Macomber A Lubbat V Distinct patterns of hippocampal pathology in Alzheimer’s disease with transactive response DNA-binding protein 43 Ann Neurol 2023 94 1036 1047 10.1002/ana.26762 37592884
Minogue G, Kawles A, Zouridakis A, Keszycki R, Macomber A, Lubbat V et al (2023) Distinct patterns of hippocampal pathology in Alzheimer’s disease with transactive response DNA-binding protein 43. Ann Neurol 94:1036–1047. 10.1002/ana.2676237592884 10.1002/ana.26762
64. Montine TJ Monsell SE Beach TG Bigio EH Bu Y Cairns NJ Multisite assessment of NIA-AA guidelines for the neuropathologic evaluation of Alzheimer’s disease Alzheimers Dement J Alzheimers Assoc 2016 12 164 169 10.1016/j.jalz.2015.07.492
Montine TJ, Monsell SE, Beach TG, Bigio EH, Bu Y, Cairns NJ et al (2016) Multisite assessment of NIA-AA guidelines for the neuropathologic evaluation of Alzheimer’s disease. Alzheimers Dement J Alzheimers Assoc 12:164–169. 10.1016/j.jalz.2015.07.49210.1016/j.jalz.2015.07.492
65. Montine TJ Phelps CH Beach TG Bigio EH Cairns NJ Dickson DW National Institute on Aging, Alzheimer’s Association National Institute on Aging-Alzheimer’s Association guidelines for the neuropathologic assessment of Alzheimer’s disease: a practical approach Acta Neuropathol (Berl) 2012 123 1 11 10.1007/s00401-011-0910-3 22101365
Montine TJ, Phelps CH, Beach TG, Bigio EH, Cairns NJ, Dickson DW et al (2012) National Institute on Aging-Alzheimer’s Association guidelines for the neuropathologic assessment of Alzheimer’s disease: a practical approach. Acta Neuropathol (Berl) 123:1–11. 10.1007/s00401-011-0910-322101365 10.1007/s00401-011-0910-3
66. Morra JH Tu Z Apostolova LG Green AE Avedissian C Madsen SK Alzheimer’s Disease Neuroimaging Initiative Automated mapping of hippocampal atrophy in 1-year repeat MRI data from 490 subjects with Alzheimer’s disease, mild cognitive impairment, and elderly controls NeuroImage 2009 45 S3 15 10.1016/j.neuroimage.2008.10.043 19041724
Morra JH, Tu Z, Apostolova LG, Green AE, Avedissian C, Madsen SK et al (2009) Automated mapping of hippocampal atrophy in 1-year repeat MRI data from 490 subjects with Alzheimer’s disease, mild cognitive impairment, and elderly controls. NeuroImage 45:S3-15. 10.1016/j.neuroimage.2008.10.04319041724 10.1016/j.neuroimage.2008.10.043
67. Nelson PT Brayne C Flanagan ME Abner EL Agrawal S Attems J Frequency of LATE neuropathologic change across the spectrum of Alzheimer’s disease neuropathology: combined data from 13 community-based or population-based autopsy cohorts Acta Neuropathol (Berl) 2022 144 27 44 10.1007/s00401-022-02444-1 35697880
Nelson PT, Brayne C, Flanagan ME, Abner EL, Agrawal S, Attems J et al (2022) Frequency of LATE neuropathologic change across the spectrum of Alzheimer’s disease neuropathology: combined data from 13 community-based or population-based autopsy cohorts. Acta Neuropathol (Berl) 144:27–44. 10.1007/s00401-022-02444-135697880 10.1007/s00401-022-02444-1
68. Nelson PT Dickson DW Trojanowski JQ Jack CR Boyle PA Arfanakis K Limbic-predominant age-related TDP-43 encephalopathy (LATE): consensus working group report Brain J Neurol 2019 142 1503 1527 10.1093/brain/awz099
Nelson PT, Dickson DW, Trojanowski JQ, Jack CR, Boyle PA, Arfanakis K et al (2019) Limbic-predominant age-related TDP-43 encephalopathy (LATE): consensus working group report. Brain J Neurol 142:1503–1527. 10.1093/brain/awz09910.1093/brain/awz099
69. Nelson PT Lee EB Cykowski MD Alafuzoff I Arfanakis K Attems J LATE-NC staging in routine neuropathologic diagnosis: an update Acta Neuropathol (Berl) 2023 145 159 173 10.1007/s00401-022-02524-2 36512061
Nelson PT, Lee EB, Cykowski MD, Alafuzoff I, Arfanakis K, Attems J et al (2023) LATE-NC staging in routine neuropathologic diagnosis: an update. Acta Neuropathol (Berl) 145:159–173. 10.1007/s00401-022-02524-236512061 10.1007/s00401-022-02524-2
70. Nelson PT Schneider JA Jicha GA Duong MT Wolk DA When Alzheimer’s is LATE: why does it matter? Ann Neurol 2023 94 211 222 10.1002/ana.26711 37245084
Nelson PT, Schneider JA, Jicha GA, Duong MT, Wolk DA (2023) When Alzheimer’s is LATE: why does it matter? Ann Neurol 94:211–222. 10.1002/ana.2671137245084 10.1002/ana.26711
71. Neltner JH Abner EL Schmitt FA Denison SK Anderson S Patel E Digital pathology and image analysis for robust high-throughput quantitative assessment of Alzheimer disease neuropathologic changes J Neuropathol Exp Neurol 2012 71 1075 1085 10.1097/NEN.0b013e3182768de4 23147505
Neltner JH, Abner EL, Schmitt FA, Denison SK, Anderson S, Patel E et al (2012) Digital pathology and image analysis for robust high-throughput quantitative assessment of Alzheimer disease neuropathologic changes. J Neuropathol Exp Neurol 71:1075–1085. 10.1097/NEN.0b013e3182768de423147505 10.1097/NEN.0b013e3182768de4
72. Neumann M Kwong LK Truax AC Vanmassenhove B Kretzschmar HA Van Deerlin VM TDP-43-positive white matter pathology in frontotemporal lobar degeneration with ubiquitin-positive inclusions J Neuropathol Exp Neurol 2007 66 177 183 10.1097/01.jnen.0000248554.45456.58 17356379
Neumann M, Kwong LK, Truax AC, Vanmassenhove B, Kretzschmar HA, Van Deerlin VM et al (2007) TDP-43-positive white matter pathology in frontotemporal lobar degeneration with ubiquitin-positive inclusions. J Neuropathol Exp Neurol 66:177–183. 10.1097/01.jnen.0000248554.45456.5817356379 10.1097/01.jnen.0000248554.45456.58
73. Neumann M Lee EB Mackenzie IR Frontotemporal lobar degeneration TDP-43-immunoreactive pathological subtypes: clinical and mechanistic significance Adv Exp Med Biol 2021 1281 201 217 10.1007/978-3-030-51140-1_13 33433877
Neumann M, Lee EB, Mackenzie IR (2021) Frontotemporal lobar degeneration TDP-43-immunoreactive pathological subtypes: clinical and mechanistic significance. Adv Exp Med Biol 1281:201–217. 10.1007/978-3-030-51140-1_1333433877 10.1007/978-3-030-51140-1_13
74. Pipitone J Park MTM Winterburn J Lett TA Lerch JP Pruessner JC Multi-atlas segmentation of the whole hippocampus and subfields using multiple automatically generated templates Neuroimage 2014 101 494 512 10.1016/j.neuroimage.2014.04.054 24784800
Pipitone J, Park MTM, Winterburn J, Lett TA, Lerch JP, Pruessner JC et al (2014) Multi-atlas segmentation of the whole hippocampus and subfields using multiple automatically generated templates. Neuroimage 101:494–512. 10.1016/j.neuroimage.2014.04.05424784800 10.1016/j.neuroimage.2014.04.054
75. Raftery AE Bayesian model selection in social research Sociol Methodol 1995 25 111 163 10.2307/271063
Raftery AE (1995) Bayesian model selection in social research. Sociol Methodol 25:111–163. 10.2307/27106310.2307/271063
76. Rani N Alm KH Corona-Long CA Speck CL Soldan A Tau PET burden in Brodmann areas 35 and 36 is associated with individual differences in cognition in non-demented older adults Front Aging Neurosci 2023 15 1272946 10.3389/fnagi.2023.1272946 38161595
Rani N, Alm KH, Corona-Long CA, Speck CL, Soldan A et al (2023) Tau PET burden in Brodmann areas 35 and 36 is associated with individual differences in cognition in non-demented older adults. Front Aging Neurosci 15:1272946. 10.3389/fnagi.2023.127294638161595 10.3389/fnagi.2023.1272946
77. Ravikumar S Wisse LEM Lim S Ittyerah R Xie L Bedard ML Ex vivo MRI atlas of the human medial temporal lobe: characterizing neurodegeneration due to tau pathology Acta Neuropathol Commun 2021 9 173 10.1186/s40478-021-01275-7 34689831
Ravikumar S, Wisse LEM, Lim S, Ittyerah R, Xie L, Bedard ML et al (2021) Ex vivo MRI atlas of the human medial temporal lobe: characterizing neurodegeneration due to tau pathology. Acta Neuropathol Commun 9:173. 10.1186/s40478-021-01275-734689831 10.1186/s40478-021-01275-7
78. Robinson JL Corrada MM Kovacs GG Dominique M Caswell C Xie SX Non-Alzheimer’s contributions to dementia and cognitive resilience in the 90+ study Acta Neuropathol (Berl) 2018 136 377 388 10.1007/s00401-018-1872-5 29916037
Robinson JL, Corrada MM, Kovacs GG, Dominique M, Caswell C, Xie SX et al (2018) Non-Alzheimer’s contributions to dementia and cognitive resilience in the 90+ study. Acta Neuropathol (Berl) 136:377–388. 10.1007/s00401-018-1872-529916037 10.1007/s00401-018-1872-5
79. Robinson JL Lee EB Xie SX Rennert L Suh E Bredenberg C Neurodegenerative disease concomitant proteinopathies are prevalent, age-related and APOE4-associated Brain 2018 141 2181 2193 10.1093/brain/awy146 29878075
Robinson JL, Lee EB, Xie SX, Rennert L, Suh E, Bredenberg C et al (2018) Neurodegenerative disease concomitant proteinopathies are prevalent, age-related and APOE4-associated. Brain 141:2181–2193. 10.1093/brain/awy14629878075 10.1093/brain/awy146
80. Schwarz CG Gunter JL Wiste HJ Przybelski SA Weigand SD Ward CP Alzheimer’s Disease Neuroimaging Initiative A large-scale comparison of cortical thickness and volume methods for measuring Alzheimer’s disease severity NeuroImage Clin 2016 11 802 812 10.1016/j.nicl.2016.05.017 28050342
Schwarz CG, Gunter JL, Wiste HJ, Przybelski SA, Weigand SD, Ward CP et al (2016) A large-scale comparison of cortical thickness and volume methods for measuring Alzheimer’s disease severity. NeuroImage Clin 11:802–812. 10.1016/j.nicl.2016.05.01728050342 10.1016/j.nicl.2016.05.017
81. Signaevsky M Prastawa M Farrell K Tabish N Baldwin E Han N Artificial intelligence in neuropathology: deep learning-based assessment of tauopathy Lab Invest 2019 99 1019 1029 10.1038/s41374-019-0202-4 30770886
Signaevsky M, Prastawa M, Farrell K, Tabish N, Baldwin E, Han N et al (2019) Artificial intelligence in neuropathology: deep learning-based assessment of tauopathy. Lab Invest 99:1019–1029. 10.1038/s41374-019-0202-430770886 10.1038/s41374-019-0202-4
82. Sims JR Zimmer JA Evans CD Lu M Ardayfio P Sparks J TRAILBLAZER-ALZ 2 Investigators Donanemab in early symptomatic Alzheimer disease: the TRAILBLAZER-ALZ 2 randomized clinical trial JAMA 2023 330 512 10.1001/jama.2023.13239 37459141
Sims JR, Zimmer JA, Evans CD, Lu M, Ardayfio P, Sparks J et al (2023) Donanemab in early symptomatic Alzheimer disease: the TRAILBLAZER-ALZ 2 randomized clinical trial. JAMA 330:512. 10.1001/jama.2023.1323937459141 10.1001/jama.2023.13239
83. Small SA Schobel SA Buxton RB Witter MP Barnes CA A pathophysiological framework of hippocampal dysfunction in ageing and disease Nat Rev Neurosci 2011 12 585 601 10.1038/nrn3085 21897434
Small SA, Schobel SA, Buxton RB, Witter MP, Barnes CA (2011) A pathophysiological framework of hippocampal dysfunction in ageing and disease. Nat Rev Neurosci 12:585–601. 10.1038/nrn308521897434 10.1038/nrn3085
84. Tang Z Chuang KV DeCarli C Jin L-W Beckett L Keiser MJ Dugger BN Interpretable classification of Alzheimer’s disease pathologies with a convolutional neural network pipeline Nat Commun 2019 10 2173 10.1038/s41467-019-10212-1 31092819
Tang Z, Chuang KV, DeCarli C, Jin L-W, Beckett L, Keiser MJ et al (2019) Interpretable classification of Alzheimer’s disease pathologies with a convolutional neural network pipeline. Nat Commun 10:2173. 10.1038/s41467-019-10212-131092819 10.1038/s41467-019-10212-1
85. Thaker AA Weinberg BD Dillon WP Hess CP Cabral HJ Fleischman DA Leurgans SE Bennett DA Hyman BT Albert MS Killiany RJ Fischl B Dale AM Desikan RS Entorhinal cortex: antemortem cortical thickness and postmortem neurofibrillary tangles and amyloid pathology Am J Neuroradiol 2017 38 961 965 10.3174/ajnr.A5133 28279988
Thaker AA, Weinberg BD, Dillon WP, Hess CP, Cabral HJ, Fleischman DA et al (2017) Entorhinal cortex: antemortem cortical thickness and postmortem neurofibrillary tangles and amyloid pathology. Am J Neuroradiol 38:961–965. 10.3174/ajnr.A513328279988 10.3174/ajnr.A5133
86. Thal DR Del Tredici K Ludolph AC Hoozemans JJM Rozemuller AJ Braak H Stages of granulovacuolar degeneration: their relation to Alzheimer’s disease and chronic stress response Acta Neuropathol (Berl) 2011 122 577 589 10.1007/s00401-011-0871-6 21935637
Thal DR, Del Tredici K, Ludolph AC, Hoozemans JJM, Rozemuller AJ, Braak H et al (2011) Stages of granulovacuolar degeneration: their relation to Alzheimer’s disease and chronic stress response. Acta Neuropathol (Berl) 122:577–589. 10.1007/s00401-011-0871-621935637 10.1007/s00401-011-0871-6
87. Thompson PM Hayashi KM De Zubicaray GI Janke AL Rose SE Semple J Mapping hippocampal and ventricular change in Alzheimer disease Neuroimage 2004 22 1754 1766 10.1016/j.neuroimage.2004.03.040 15275931
Thompson PM, Hayashi KM, De Zubicaray GI, Janke AL, Rose SE, Semple J et al (2004) Mapping hippocampal and ventricular change in Alzheimer disease. Neuroimage 22:1754–1766. 10.1016/j.neuroimage.2004.03.04015275931 10.1016/j.neuroimage.2004.03.040
88. Toledo JB Van Deerlin VM Lee EB Suh E Baek Y Robinson JL A platform for discovery: the University of Pennsylvania integrated neurodegenerative disease biobank Alzheimers Dement J Alzheimers Assoc 2014 10 477 484.e1 10.1016/j.jalz.2013.06.003
Toledo JB, Van Deerlin VM, Lee EB, Suh E, Baek Y, Robinson JL et al (2014) A platform for discovery: the University of Pennsylvania integrated neurodegenerative disease biobank. Alzheimers Dement J Alzheimers Assoc 10:477-484.e1. 10.1016/j.jalz.2013.06.00310.1016/j.jalz.2013.06.003
89. Uchihara T Pretangles and neurofibrillary changes: similarities and differences between AD and CBD based on molecular and morphological evolution Neuropathol Off J Jpn Soc Neuropathol 2014 34 571 577 10.1111/neup.12108
Uchihara T (2014) Pretangles and neurofibrillary changes: similarities and differences between AD and CBD based on molecular and morphological evolution. Neuropathol Off J Jpn Soc Neuropathol 34:571–577. 10.1111/neup.1210810.1111/neup.12108
90. Vizcarra JC Pearce TM Dugger BN Keiser MJ Gearing M Crary JF Toward a generalizable machine learning workflow for neurodegenerative disease staging with focus on neurofibrillary tangles Acta Neuropathol Commun 2023 11 202 10.1186/s40478-023-01691-x 38110981
Vizcarra JC, Pearce TM, Dugger BN, Keiser MJ, Gearing M, Crary JF et al (2023) Toward a generalizable machine learning workflow for neurodegenerative disease staging with focus on neurofibrillary tangles. Acta Neuropathol Commun 11:202. 10.1186/s40478-023-01691-x38110981 10.1186/s40478-023-01691-x
91. Walker L McAleese KE Johnson M Khundakar AA Erskine D Thomas AJ Quantitative neuropathology: an update on automated methodologies and implications for large scale cohorts J Neural Transm Vienna Austria. 2017 124 671 683 10.1007/s00702-017-1702-2
Walker L, McAleese KE, Johnson M, Khundakar AA, Erskine D, Thomas AJ et al (2017) Quantitative neuropathology: an update on automated methodologies and implications for large scale cohorts. J Neural Transm Vienna Austria. 124:671–683. 10.1007/s00702-017-1702-210.1007/s00702-017-1702-2
92. Walker L McAleese KE Thomas AJ Johnson M Martin-Ruiz C Parker C Neuropathologically mixed Alzheimer’s and Lewy body disease: burden of pathological protein aggregates differs between clinical phenotypes Acta Neuropathol (Berl) 2015 129 729 748 10.1007/s00401-015-1406-3 25758940
Walker L, McAleese KE, Thomas AJ, Johnson M, Martin-Ruiz C, Parker C et al (2015) Neuropathologically mixed Alzheimer’s and Lewy body disease: burden of pathological protein aggregates differs between clinical phenotypes. Acta Neuropathol (Berl) 129:729–748. 10.1007/s00401-015-1406-325758940 10.1007/s00401-015-1406-3
93. Wennberg AM Whitwell JL Tosakulwong N Weigand SD Murray ME Machulda MM The influence of tau, amyloid, alpha-synuclein, TDP-43, and vascular pathology in clinically normal elderly individuals Neurobiol Aging 2019 77 26 36 10.1016/j.neurobiolaging.2019.01.008 30776649
Wennberg AM, Whitwell JL, Tosakulwong N, Weigand SD, Murray ME, Machulda MM et al (2019) The influence of tau, amyloid, alpha-synuclein, TDP-43, and vascular pathology in clinically normal elderly individuals. Neurobiol Aging 77:26–36. 10.1016/j.neurobiolaging.2019.01.00830776649 10.1016/j.neurobiolaging.2019.01.008
94. Wisse LEM (2022) T1-ASHS Quality Control. https://www.youtube.com/watch?v=Hb4X8Q_bWo4. Accessed 24 Apr 2024
95. Wisse LEM de Flores R Xie L Das SR McMillan CT Trojanowski JQ On behalf of the Alzheimer’s Disease NeuroImaging Initiative Pathological drivers of neurodegeneration in suspected non-Alzheimer’s disease pathophysiology Alzheimers Res Ther 2021 13 100 10.1186/s13195-021-00835-2 33990226
Wisse LEM, de Flores R, Xie L, Das SR, McMillan CT, Trojanowski JQ et al (2021) Pathological drivers of neurodegeneration in suspected non-Alzheimer’s disease pathophysiology. Alzheimers Res Ther 13:100. 10.1186/s13195-021-00835-233990226 10.1186/s13195-021-00835-2
96. Wisse LEM Ravikumar S Ittyerah R Lim S Lane J Bedard ML Downstream effects of polypathology on neurodegeneration of medial temporal lobe subregions Acta Neuropathol Commun 2021 9 128 10.1186/s40478-021-01225-3 34289895
Wisse LEM, Ravikumar S, Ittyerah R, Lim S, Lane J, Bedard ML et al (2021) Downstream effects of polypathology on neurodegeneration of medial temporal lobe subregions. Acta Neuropathol Commun 9:128. 10.1186/s40478-021-01225-334289895 10.1186/s40478-021-01225-3
97. Wolk DA Das SR Mueller SG Weiner MW Yushkevich PA Medial temporal lobe subregional morphometry using high resolution MRI in Alzheimer’s disease Neurobiol Aging 2017 49 204 213 10.1016/j.neurobiolaging.2016.09.011 27836336
Wolk DA, Das SR, Mueller SG, Weiner MW, Yushkevich PA (2017) Medial temporal lobe subregional morphometry using high resolution MRI in Alzheimer’s disease. Neurobiol Aging 49:204–213. 10.1016/j.neurobiolaging.2016.09.01127836336 10.1016/j.neurobiolaging.2016.09.011
98. Wuestefeld A Pichet Binette A Berron D Spotorno N van Westen D Stomrud E Age-related and amyloid-beta-independent tau deposition and its downstream effects Brain 2023 146 3192 3205 10.1093/brain/awad135 37082959
Wuestefeld A, Pichet Binette A, Berron D, Spotorno N, van Westen D, Stomrud E et al (2023) Age-related and amyloid-beta-independent tau deposition and its downstream effects. Brain 146:3192–3205. 10.1093/brain/awad13537082959 10.1093/brain/awad135
99. Wurts A Oakley DH Hyman BT Samsi S Segmentation of Tau stained alzheimers brain tissue using convolutional neural networks Annu Int Conf IEEE Eng Med Biol Soc IEEE Eng Med Biol Soc Annu Int Conf 2020 2020 1420 1423 10.1109/EMBC44109.2020.9175832
Wurts A, Oakley DH, Hyman BT, Samsi S (2020) Segmentation of Tau stained alzheimers brain tissue using convolutional neural networks. Annu Int Conf IEEE Eng Med Biol Soc IEEE Eng Med Biol Soc Annu Int Conf 2020:1420–1423. 10.1109/EMBC44109.2020.917583210.1109/EMBC44109.2020.9175832
100. Xie L Wisse LEM Das SR Vergnet N Dong M Ittyerah R For the Alzheimer’s Disease Neuroimaging Initiative Longitudinal atrophy in early Braak regions in preclinical Alzheimer’s disease Hum Brain Mapp 2020 41 4704 4717 10.1002/hbm.25151 32845545
Xie L, Wisse LEM, Das SR, Vergnet N, Dong M, Ittyerah R et al (2020) Longitudinal atrophy in early Braak regions in preclinical Alzheimer’s disease. Hum Brain Mapp 41:4704–4717. 10.1002/hbm.2515132845545 10.1002/hbm.25151
101. Xie L Wisse LEM Pluta J de Flores R Piskin V Manjón JV Automated segmentation of medial temporal lobe subregions on in vivo T1-weighted MRI in early stages of Alzheimer’s disease Hum Brain Mapp 2019 40 3431 3451 10.1002/hbm.24607 31034738
Xie L, Wisse LEM, Pluta J, de Flores R, Piskin V, Manjón JV et al (2019) Automated segmentation of medial temporal lobe subregions on in vivo T1-weighted MRI in early stages of Alzheimer’s disease. Hum Brain Mapp 40:3431–3451. 10.1002/hbm.2460731034738 10.1002/hbm.24607
102. Yu L Boyle PA Dawe RJ Bennett DA Arfanakis K Schneider JA Contribution of TDP and hippocampal sclerosis to hippocampal volume loss in older-old persons Neurology 2020 94 e142 e152 10.1212/WNL.0000000000008679 31757868
Yu L, Boyle PA, Dawe RJ, Bennett DA, Arfanakis K, Schneider JA (2020) Contribution of TDP and hippocampal sclerosis to hippocampal volume loss in older-old persons. Neurology 94:e142–e152. 10.1212/WNL.000000000000867931757868 10.1212/WNL.0000000000008679
103. Yushkevich PA MuñozLópez M Iñiguez de OnzoñoMartin MM Ittyerah R Lim S Three-dimensional mapping of neurofibrillary tangle burden in the human medial temporal lobe Brain 2021 144 2784 2797 10.1093/brain/awab262 34259858
Yushkevich PA, MuñozLópez M, Iñiguez de OnzoñoMartin MM, Ittyerah R, Lim S et al (2021) Three-dimensional mapping of neurofibrillary tangle burden in the human medial temporal lobe. Brain 144:2784–2797. 10.1093/brain/awab26234259858 10.1093/brain/awab262
104. Yushkevich PA Pluta JB Wang H Xie L Ding S-L Gertje EC Automated volumetry and regional thickness analysis of hippocampal subfields and medial temporal cortical structures in mild cognitive impairment Hum Brain Mapp 2015 36 258 287 10.1002/hbm.22627 25181316
Yushkevich PA, Pluta JB, Wang H, Xie L, Ding S-L, Gertje EC et al (2015) Automated volumetry and regional thickness analysis of hippocampal subfields and medial temporal cortical structures in mild cognitive impairment. Hum Brain Mapp 36:258–287. 10.1002/hbm.2262725181316 10.1002/hbm.22627
105. Yushkevich PA Xie L Wisse LE Dong M Ravikumar S Ittyerah R de Flores R Das SR Wolk DA Mapping medial temporal lobe longitudinal change in preclinical Alzheimer’s disease Alzheimers Dement 2023 19 e081898 10.1002/alz.081898
Yushkevich PA, Xie L, Wisse LE, Dong M, Ravikumar S, Ittyerah R et al (2023) Mapping medial temporal lobe longitudinal change in preclinical Alzheimer’s disease. Alzheimers Dement 19:e081898. 10.1002/alz.08189810.1002/alz.081898
