Deep learning assisted quantitative assessment of histopathological markers of Alzheimer's disease and cerebral amyloid angiopathy

Neuropathological Correlates of White Matter Hyperintensities in Cerebral Amyloid Angiopathy

BACKGROUND

White matter hyperintensities (WMHs) are frequently observed on magnetic resonance imaging (MRI) in patients with cerebral amyloid angiopathy (CAA). The neuropathological substrates that underlie WMHs in CAA are unclear, and it remains largely unexplored whether the different WMH distribution patterns associated with CAA (posterior confluent and subcortical multispot) reflect alternative pathophysiological mechanisms.

METHODS AND RESULTS

We performed a combined in vivo MRI-ex vivo MRI-neuropathological study in patients with definite CAA. Formalin-fixed hemispheres from 19 patients with CAA, most of whom also had in vivo MRI available, underwent 3T MRI, followed by standard neuropathological examination of the hemispheres and targeted neuropathological assessment of WMH patterns. Ex vivo WMH volume was independently associated with CAA severity (P=0.046) but not with arteriolosclerosis (P=0.743). In targeted neuropathological examination, compared with normal-appearing white matter, posterior confluent WMHs were associated with activated microglia (P=0.043) and clasmatodendrosis (P=0.031), a form of astrocytic injury. Trends were found for an association with white matter rarefaction (P=0.074) and arteriolosclerosis (P=0.094). An exploratory descriptive analysis suggested that the histopathological correlates of WMH multispots were similar to those underlying posterior confluent WMHs.

CONCLUSIONS

This study confirmed that vascular amyloid β severity in the cortex is significantly associated with WMH volume in patients with definite CAA. The histopathological substrates of both posterior confluent and WMH multispots were comparable, suggesting overlapping pathophysiological mechanisms, although these exploratory observations require confirmation in larger studies.

Deep learning assisted quantitative analysis of Aβ and microglia in patients with idiopathic normal pressure hydrocephalus in relation to cognitive outcome

Neuropathologic changes of Alzheimer disease (AD) including Aβ accumulation and neuroinflammation are frequently observed in the cerebral cortex of patients with idiopathic normal pressure hydrocephalus (iNPH). We created an automated analysis platform to quantify Aβ load and reactive microglia in the vicinity of Aβ plaques and to evaluate their association with cognitive outcome in cortical biopsies of patients with iNPH obtained at the time of shunting. Aiforia Create deep learning software was used on whole slide images of Iba1/4G8 double immunostained frontal cortical biopsies of 120 shunted iNPH patients to identify Iba1-positive microglia somas and Aβ areas, respectively. Dementia, AD clinical syndrome (ACS), and Clinical Dementia Rating Global score (CDR-GS) were evaluated retrospectively after a median follow-up of 4.4 years. Deep learning artificial intelligence yielded excellent (>95%) precision for tissue, Aβ, and microglia somas. Using an age-adjusted model, higher Aβ coverage predicted the development of dementia, the diagnosis of ACS, and more severe memory impairment by CDR-GS whereas measured microglial densities and Aβ-related microglia did not correlate with cognitive outcome in these patients. Therefore, cognitive outcome seems to be hampered by higher Aβ coverage in cortical biopsies in shunted iNPH patients but is not correlated with densities of surrounding microglia.

Sex differences in histopathological markers of cerebral amyloid angiopathy and related hemorrhage

BACKGROUND

Men with cerebral amyloid angiopathy (CAA) may have an earlier onset of intracerebral hemorrhage and a more hemorrhagic disease course compared to women. In this cohort study, we investigated sex differences in histopathological markers associated with amyloid-β burden and hemorrhage in cognitively impaired individuals and patients with CAA, using neuropathological data from two autopsy databases.

METHODS

First, we investigated presence of parenchymal (Thal score) and vascular amyloid-β (CAA severity score) in cognitively impaired individuals from the National Alzheimer's Coordinating Center (NACC) neuropathology database. Next, we examined sex differences in hemorrhagic ex vivo magnetic resonance imaging (MRI) markers and local cortical iron burden and the interaction of sex on factors associated with cortical iron burden (CAA percentage area and vessel remodeling) in patients with pathologically confirmed clinical CAA from the Massachusetts General Hospital (MGH) CAA neuropathology database.

RESULTS

In 6120 individuals from the NACC database (45% women, mean age 80 years), the presence of parenchymal amyloid-β (odds ratio (OR) (95% confidence interval (CI)) =0.68 (0.53-0.88)) but not vascular amyloid-β was less in men compared to women. In 19 patients with definite CAA from the MGH CAA database (35% women, mean age 75 years), a lower microbleed count (p < 0.001) but a higher proportion of cortical superficial siderosis and a higher local cortical iron burden was found in men (p < 0.001) compared to women. CAA percentage area was comparable in men and women (p = 0.732). Exploratory analyses demonstrated a possible stronger negative relation between cortical CAA percentage area and cortical iron density in men compared to women (p = 0.03).

CONCLUSION

Previously observed sex differences in hemorrhage onset and progression in CAA patients are likely not due to differences in global CAA severity between men and women. Other factors, such as vascular remodeling, may contribute, but future studies are necessary to replicate our findings in larger data sets and to further investigate the underlying mechanisms behind these complex sex differences.

Neuron collinearity differentiates human hippocampal subregions: a validated deep learning approach

The hippocampus is heterogeneous in its architecture. It contributes to cognitive processes such as memory and spatial navigation and is susceptible to neurodegenerative disease. Cytoarchitectural features such as neuron size and neuronal collinearity have been used to parcellate the hippocampal subregions. Moreover, pyramidal neuron orientation (orientation of one individual neuron) and collinearity (how neurons align) have been investigated as a measure of disease in schizophrenia. However, a comprehensive quantitative study of pyramidal neuron orientation and collinearity within the hippocampal subregions has not yet been conducted. In this study, we present a high-throughput deep learning approach for the automated extraction of pyramidal neuron orientation in the hippocampal subregions. Based on the pretrained Cellpose algorithm for cellular segmentation, we measured 479 873 pyramidal neurons in 168 hippocampal partitions. We corrected the neuron orientation estimates to account for the curvature of the hippocampus and generated collinearity measures suitable for inter- and intra-individual comparisons. Our deep learning results were validated with manual orientation assessment. This study presents a quantitative metric of pyramidal neuron collinearity within the hippocampus. It reveals significant differences among the individual hippocampal subregions (P  < 0.001), with cornu ammonis 3 being the most collinear, followed by cornu ammonis 2, cornu ammonis 1, the medial/uncal subregions and subiculum. Our data establishes pyramidal neuron collinearity as a quantitative parameter for hippocampal subregion segmentation, including the differentiation of cornu ammonis 2 and cornu ammonis 3. This novel deep learning approach could facilitate large-scale multicentric analyses in subregion parcellation and lays groundwork for the investigation of mental illnesses at the cellular level.

Three-dimensional histology reveals dissociable human hippocampal long axis gradients of Alzheimer's pathology

INTRODUCTION

Three-dimensional (3D) histology analyses are essential to overcome sampling variability and understand pathological differences beyond the dissection axis. We present Path2MR, the first pipeline allowing 3D reconstruction of sparse human histology without an MRI reference. We implemented Path2MR with post-mortem hippocampal sections to explore pathology gradients in Alzheimer's Disease.

METHODS

Blockface photographs of brain hemisphere slices are used for 3D reconstruction, from which an MRI-like image is generated using machine learning. Histology sections are aligned to the reconstructed hemisphere and subsequently to an atlas in standard space.

RESULTS

Path2MR successfully registered histological sections to their anatomical position along the hippocampal longitudinal axis. Combined with histopathology quantification, we found an expected peak of tau pathology at the anterior end of the hippocampus, while amyloid-β displayed a quadratic anterior-posterior distribution.

CONCLUSION

Path2MR, which enables 3D histology using any brain bank dataset, revealed significant differences along the hippocampus between tau and amyloid-β.

High-throughput digital quantification of Alzheimer disease pathology and associated infrastructure in large autopsy studies

High-throughput digital pathology offers considerable advantages over traditional semiquantitative and manual methods of counting pathology. We used brain tissue from 5 clinical-pathologic cohort studies of aging; the Religious Orders Study, the Rush Memory and Aging Project, the Minority Aging Research Study, the African American Clinical Core, and the Latino Core to (1) develop a workflow management system for digital pathology processes, (2) optimize digital algorithms to quantify Alzheimer disease (AD) pathology, and (3) harmonize data statistically. Data from digital algorithms for the quantification of β-amyloid (Aβ, n = 413) whole slide images and tau-tangles (n = 639) were highly correlated with manual pathology data (r = 0.83 to 0.94). Measures were robust and reproducible across different magnifications and repeated scans. Digital measures for Aβ and tau-tangles across multiple brain regions reproduced established patterns of correlations, even when samples were stratified by clinical diagnosis. Finally, we harmonized newly generated digital measures with historical measures across multiple large autopsy-based studies. We describe a multidisciplinary approach to develop a digital pathology pipeline that reproducibly identifies AD neuropathologies, Aβ load, and tau-tangles. Digital pathology is a powerful tool that can overcome critical challenges associated with traditional microscopy methods.

Cortical superficial siderosis is associated with reactive astrogliosis in cerebral amyloid angiopathy

BACKGROUND

Cortical superficial siderosis (cSS) has recently emerged as one of the most important predictors of symptomatic intracerebral hemorrhage and is a risk factor for post-stroke dementia in cerebral amyloid angiopathy (CAA). However, it remains unknown whether cSS is just a marker of severe CAA pathology or may itself contribute to intracerebral hemorrhage risk and cognitive decline. cSS is a chronic manifestation of convexal subarachnoid hemorrhage and is neuropathologically characterized by iron deposits in the superficial cortical layers. We hypothesized that these iron deposits lead to local neuroinflammation, a potentially contributory pathway towards secondary tissue injury.

METHODS

Accordingly, we assessed the distribution of inflammatory markers in relation to cortical iron deposits in post-mortem tissue from CAA cases. Serial sections from the frontal, parietal, temporal, and occipital lobes of nineteen autopsy cases with CAA were stained with Perls' Prussian blue (iron) and underwent immunohistochemistry against glial fibrillary acidic protein (GFAP, reactive astrocytes) and cluster of differentiation 68 (CD68, activated microglia/macrophages). Digitized sections were uploaded to the cloud-based Aiforia® platform, where deep-learning algorithms were utilized to detect tissue, iron deposits, and GFAP-positive and CD68-positive cells.

RESULTS

We observed a strong local relationship between cortical iron deposits and reactive astrocytes. Like cSS-related iron, reactive astrocytes were mainly found in the most superficial layers of the cortex. Although we observed iron within both astrocytes and activated microglia/macrophages on co-stains, there was no clear local relationship between the density of microglia/macrophages and the density of iron deposits.

CONCLUSION

Iron deposition resulting from cSS is associated with local reactive astrogliosis.

Learning fast and fine-grained detection of amyloid neuropathologies from coarse-grained expert labels

Precise, scalable, and quantitative evaluation of whole slide images is crucial in neuropathology. We release a deep learning model for rapid object detection and precise information on the identification, locality, and counts of cored plaques and cerebral amyloid angiopathy (CAA). We trained this object detector using a repurposed image-tile dataset without any human-drawn bounding boxes. We evaluated the detector on a new manually-annotated dataset of whole slide images (WSIs) from three institutions, four staining procedures, and four human experts. The detector matched the cohort of neuropathology experts, achieving 0.64 (model) vs. 0.64 (cohort) average precision (AP) for cored plaques and 0.75 vs. 0.51 AP for CAAs at a 0.5 IOU threshold. It provided count and locality predictions that approximately correlated with gold-standard human CERAD-like WSI scoring (p = 0.07 ± 0.10). The openly-available model can quickly score WSIs in minutes without a GPU on a standard workstation.

Stereology neuron counts correlate with deep learning estimates in the human hippocampal subregions

Hippocampal subregions differ in specialization and vulnerability to cell death. Neuron death and hippocampal atrophy have been a marker for the progression of Alzheimer's disease. Relatively few studies have examined neuronal loss in the human brain using stereology. We characterize an automated high-throughput deep learning pipeline to segment hippocampal pyramidal neurons, generate pyramidal neuron estimates within the human hippocampal subfields, and relate our results to stereology neuron counts. Based on seven cases and 168 partitions, we vet deep learning parameters to segment hippocampal pyramidal neurons from the background using the open-source CellPose algorithm, and show the automated removal of false-positive segmentations. There was no difference in Dice scores between neurons segmented by the deep learning pipeline and manual segmentations (Independent Samples t-Test: t(28) = 0.33, p = 0.742). Deep-learning neuron estimates strongly correlate with manual stereological counts per subregion (Spearman's correlation (n = 9): r(7) = 0.97, p < 0.001), and for each partition individually (Spearman's correlation (n = 168): r(166) = 0.90, p <0 .001). The high-throughput deep-learning pipeline provides validation to existing standards. This deep learning approach may benefit future studies in tracking baseline and resilient healthy aging to the earliest disease progression.

Learning fast and fine-grained detection of amyloid neuropathologies from coarse-grained expert labels

Precise, scalable, and quantitative evaluation of whole slide images is crucial in neuropathology. We release a deep learning model for rapid object detection and precise information on the identification, locality, and counts of cored plaques and cerebral amyloid angiopathies (CAAs). We trained this object detector using a repurposed image-tile dataset without any human-drawn bounding boxes. We evaluated the detector on a new manually-annotated dataset of whole slide images (WSIs) from three institutions, four staining procedures, and four human experts. The detector matched the cohort of neuropathology experts, achieving 0.64 (model) vs. 0.64 (cohort) average precision (AP) for cored plaques and 0.75 vs. 0.51 AP for CAAs at a 0.5 IOU threshold. It provided count and locality predictions that correlated with gold-standard CERAD-like WSI scoring (p=0.07± 0.10). The openly-available model can quickly score WSIs in minutes without a GPU on a standard workstation.

Deep learning from multiple experts improves identification of amyloid neuropathologies

Pathologists can label pathologies differently, making it challenging to yield consistent assessments in the absence of one ground truth. To address this problem, we present a deep learning (DL) approach that draws on a cohort of experts, weighs each contribution, and is robust to noisy labels. We collected 100,495 annotations on 20,099 candidate amyloid beta neuropathologies (cerebral amyloid angiopathy (CAA), and cored and diffuse plaques) from three institutions, independently annotated by five experts. DL methods trained on a consensus-of-two strategy yielded 12.6-26% improvements by area under the precision recall curve (AUPRC) when compared to those that learned individualized annotations. This strategy surpassed individual-expert models, even when unfairly assessed on benchmarks favoring them. Moreover, ensembling over individual models was robust to hidden random annotators. In blind prospective tests of 52,555 subsequent expert-annotated images, the models labeled pathologies like their human counterparts (consensus model AUPRC = 0.74 cored; 0.69 CAA). This study demonstrates a means to combine multiple ground truths into a common-ground DL model that yields consistent diagnoses informed by multiple and potentially variable expert opinions.

Perivascular space dilation is associated with vascular amyloid-β accumulation in the overlying cortex

Perivascular spaces (PVS) are compartments surrounding cerebral blood vessels that become visible on MRI when enlarged. Enlarged PVS (EPVS) are commonly seen in patients with cerebral small vessel disease (CSVD) and have been suggested to reflect dysfunctional perivascular clearance of soluble waste products from the brain. In this study, we investigated histopathological correlates of EPVS and how they relate to vascular amyloid-β (Aβ) in cerebral amyloid angiopathy (CAA), a form of CSVD that commonly co-exists with Alzheimer's disease (AD) pathology. We used ex vivo MRI, semi-automatic segmentation and validated deep-learning-based models to quantify EPVS and associated histopathological abnormalities. Severity of MRI-visible PVS during life was significantly associated with severity of MRI-visible PVS on ex vivo MRI in formalin fixed intact hemispheres and corresponded with PVS enlargement on histopathology in the same areas. EPVS were located mainly around the white matter portion of perforating cortical arterioles and their burden was associated with CAA severity in the overlying cortex. Furthermore, we observed markedly reduced smooth muscle cells and increased vascular Aβ accumulation, extending into the WM, in individually affected vessels with an EPVS. Overall, these findings are consistent with the notion that EPVS reflect impaired outward flow along arterioles and have implications for our understanding of perivascular clearance mechanisms, which play an important role in the pathophysiology of CAA and AD.

Histopathological correlates of haemorrhagic lesions on ex vivo magnetic resonance imaging in immunized Alzheimer's disease cases

Haemorrhagic amyloid-related imaging abnormalities on MRI are frequently observed adverse events in the context of amyloid β immunotherapy trials in patients with Alzheimer's disease. The underlying histopathology and pathophysiological mechanisms of haemorrhagic amyloid-related imaging abnormalities remain largely unknown, although coexisting cerebral amyloid angiopathy may play a key role. Here, we used ex vivo MRI in cases that underwent amyloid β immunotherapy during life to screen for haemorrhagic lesions and assess underlying tissue and vascular alterations. We hypothesized that these lesions would be associated with severe cerebral amyloid angiopathy. Ten cases were selected from the long-term follow-up study of patients who enrolled in the first clinical trial of active amyloid β immunization with AN1792 for Alzheimer's disease. Eleven matched non-immunized Alzheimer's disease cases from an independent brain brank were used as 'controls'. Formalin-fixed occipital brain slices were imaged at 7 T MRI to screen for haemorrhagic lesions (i.e. microbleeds and cortical superficial siderosis). Samples with and without haemorrhagic lesions were cut and stained. Artificial intelligence-assisted quantification of amyloid β plaque area, cortical and leptomeningeal cerebral amyloid angiopathy area, the density of iron and calcium positive cells and reactive astrocytes and activated microglia was performed. On ex vivo MRI, cortical superficial siderosis was observed in 5/10 immunized Alzheimer's disease cases compared with 1/11 control Alzheimer's disease cases (κ = 0.5). On histopathology, these areas revealed iron and calcium positive deposits in the cortex. Within the immunized Alzheimer's disease group, areas with siderosis on MRI revealed greater leptomeningeal cerebral amyloid angiopathy and concentric splitting of the vessel walls compared with areas without siderosis. Moreover, greater density of iron-positive cells in the cortex was associated with lower amyloid β plaque area and a trend towards increased post-vaccination antibody titres. This work highlights the use of ex vivo MRI to investigate the neuropathological correlates of haemorrhagic lesions observed in the context of amyloid β immunotherapy. These findings suggest a possible role for cerebral amyloid angiopathy in the formation of haemorrhagic amyloid-related imaging abnormalities, awaiting confirmation in future studies that include brain tissue of patients who received passive immunotherapy against amyloid β with available in vivo MRI during life.