Diagnostic Accuracy of a Custom Large Language Model on Rare Pediatric Disease Case Reports.

Accurately diagnosing rare pediatric diseases frequently represent a clinical challenge due to their complex and unusual clinical presentations. Here, we explore the capabilities of three large language models (LLMs), GPT-4, Gemini Pro, and a custom-built LLM (GPT-4 integrated with the Human Phenotype Ontology [GPT-4 HPO]), by evaluating their diagnostic performance on 61 rare pediatric disease case reports. The performance of the LLMs were assessed for accuracy in identifying specific diagnoses, listing the correct diagnosis among a differential list, and broad disease categories. In addition, GPT-4 HPO was tested on 100 general pediatrics case reports previously assessed on other LLMs to further validate its performance. The results indicated that GPT-4 was able to predict the correct diagnosis with a diagnostic accuracy of 13.1%, whereas both GPT-4 HPO and Gemini Pro had diagnostic accuracies of 8.2%. Further, GPT-4 HPO showed an improved performance compared with the other two LLMs in identifying the correct diagnosis among its differential list and the broad disease category. Although these findings underscore the potential of LLMs for diagnostic support, particularly when enhanced with domain-specific ontologies, they also stress the need for further improvement prior to integration into clinical practice.

Loss of spontaneous vasomotion precedes impaired cerebrovascular reactivity and microbleeds in a mouse model of cerebral amyloid angiopathy.

Background: Cerebral amyloid angiopathy (CAA) is a cerebral small vessel disease in which amyloid-ß accumulates in vessel walls. CAA is a leading cause of symptomatic lobar intracerebral hemorrhage and an important contributor to age-related cognitive decline. Recent work has suggested that vascular dysfunction may precede symptomatic stages of CAA, and that spontaneous slow oscillations in arteriolar diameter (termed vasomotion), important for amyloid-ß clearance, may be impaired in CAA. Methods: To systematically study the progression of vascular dysfunction in CAA, we used the APP23 mouse model of amyloidosis, which is known to develop spontaneous cerebral microbleeds mimicking human CAA. Using in vivo 2-photon microscopy, we longitudinally imaged unanesthetized APP23 transgenic mice and wildtype littermates from 7 to 14 months of age, tracking amyloid-ß accumulation and vasomotion in individual pial arterioles over time. MRI was used in separate groups of 12-, 18-, and 24-month-old APP23 transgenic mice and wildtype littermates to detect microbleeds and to assess cerebral blood flow and cerebrovascular reactivity with pseudo-continuous arterial spin labeling. Results: We observed a significant decline in vasomotion with age in APP23 mice, while vasomotion remained unchanged in wildtype mice with age. This decline corresponded in timing to initial vascular amyloid-ß deposition (∼8-10 months of age), although was more strongly correlated with age than with vascular amyloid-ß burden in individual arterioles. Declines in vasomotion preceded the development of MRI-visible microbleeds and the loss of smooth muscle actin in arterioles, both of which were observed in APP23 mice by 18 months of age. Additionally, evoked cerebrovascular reactivity was intact in APP23 mice at 12 months of age, but significantly lower in APP23 mice by 24 months of age. Conclusions: Our findings suggest that a decline in spontaneous vasomotion is an early, potentially pre-symptomatic, manifestation of CAA and vascular dysfunction, and a possible future treatment target.

Age-Dependent Increase in Tau Phosphorylation at Serine 396 in Huntington's Disease Prefrontal Cortex.

BACKGROUND: To date, it is still controversial whether tau phosphorylation plays a role in Huntington's disease (HD), as previous studies demonstrated either no alterations or increases in phosphorylated tau (pTau) in HD postmortem brain and mouse models. OBJECTIVE: The goal of this study was to determine whether total tau and pTau levels are altered in HD. METHODS: Immunohistochemistry, cellular fractionations, and western blots were used to measure total tau and pTau levels in a large cohort of HD and control postmortem prefrontal cortex (PFC). Furthermore, western blots were performed to assess tau, and pTau levels in HD and control isogenic embryonic stem cell (ESC)-derived cortical neurons and neuronal stem cells (NSCs). Similarly, western blots were used to assess tau and pTau levels in HttQ111 and transgenic R6/2 mice. Lastly, total tau levels were assessed in HD and healthy control plasma using Quanterix Simoa assay. RESULTS: Our results revealed that, while there was no difference in total tau or pTau levels in HD PFC compared to controls, the levels of tau phosphorylated at S396 were increased in PFC samples from HD patients 60 years or older at time of death. Additionally, tau and pTau levels were not changed in HD ESC-derived cortical neurons and NSCs. Similarly, total tau or pTau levels were not altered in HttQ111 and transgenic R6/2 mice compared to wild-type littermates. Lastly, tau levels were not changed in plasma from a small cohort of HD patients compared to controls. CONCLUSIONS: Together these findings demonstrate that pTau-S396 levels increase significantly with age in HD PFC.

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.

Histopathological Correlates of Lobar Microbleeds in False-Positive Cerebral Amyloid Angiopathy Cases.

OBJECTIVE: A definite diagnosis of cerebral amyloid angiopathy (CAA), characterized by the accumulation of amyloid ß in walls of cerebral small vessels, can only be obtained through pathological examination. A diagnosis of probable CAA during life relies on the presence of hemorrhagic markers, including lobar cerebral microbleeds (CMBs). The aim of this project was to study the histopathological correlates of lobar CMBs in false-positive CAA cases. METHODS: In 3 patients who met criteria for probable CAA during life, but showed no CAA upon neuropathological examination, lobar CMBs were counted on ex vivo 3T magnetic resonance imaging (MRI) and on ex vivo 7T MRI. Areas with lobar CMBs were next sampled and cut into serial sections, on which the CMBs were then identified. RESULTS: Collectively, there were 25 lobar CMBs on in vivo MRI and 22 on ex vivo 3T MRI of the analyzed hemispheres. On ex vivo MRI, we targeted 12 CMBs for sampling, and definite histopathological correlates were retrieved for 9 of them, of which 7 were true CMBs. No CAA was found on any of the serial sections. The "culprit vessels" associated with the true CMBs instead showed moderate to severe arteriolosclerosis. Furthermore, CMBs in false-positive CAA cases tended to be located more often in the juxtacortical or subcortical white matter than in the cortical ribbon. INTERPRETATION: These findings suggest that arteriolosclerosis can generate lobar CMBs and that more detailed investigations into the exact localization of CMBs with respect to the cortical ribbon could potentially aid the diagnosis of CAA during life. ANN NEUROL 2023;94:856-870.

Age-dependent increase in tau phosphorylation at serine 396 in Huntington's disease pre-frontal cortex.

Background: To date, it is still controversial whether tau phosphorylation plays a role in Huntington's disease (HD), as previous studies demonstrated either no alterations or increases in phosphorylated tau (pTau) in HD post-mortem brain and mouse models. Objectives: The goal of this study was to determine whether total tau and pTau levels are altered in HD. Methods: Immunohistochemistry, cellular fractionations, and western blots were used to measure tau and pTau levels in a large cohort of HD and control post-mortem prefrontal cortex (PFC). Furthermore, western blots were performed to assess tau, and pTau levels in HD and control isogenic embryonic stem cell (ESC)-derived cortical neurons and neuronal stem cells (NSCs). Similarly, western blots were used to assess tau and pTau in Htt Q111 and transgenic R6/2 mice. Lastly, total tau levels were assessed in HD and healthy control plasma using Quanterix Simoa assay. Results: Our results revealed that, while there was no difference in tau or pTau levels in HD PFC compared to controls, tau phosphorylated at S396 levels were increased in PFC samples from HD patients 60 years or older at time of death. Additionally, tau and pTau levels were not changed in HD ESC-derived cortical neurons and NSCs. Similarly, tau or pTau levels were not altered in Htt Q111 and transgenic R6/2 mice compared to wild-type littermates. Lastly, tau levels were not changed in plasma from a small cohort of HD patients compared to controls. Conclusion: Together these findings demonstrate that pTau-S396 levels increase significantly with age in HD PFC.

Spontaneous vasomotion propagates along pial arterioles in the awake mouse brain like stimulus-evoked vascular reactivity.

Sensory stimulation evokes a local, vasodilation-mediated blood flow increase to the activated brain region, which is referred to as functional hyperemia. Spontaneous vasomotion is a change in arteriolar diameter that occurs without sensory stimulation, at low frequency (∼0.1 Hz). These vessel diameter changes are a driving force for perivascular soluble waste clearance, the failure of which has been implicated in neurodegenerative disease. Stimulus-evoked vascular reactivity is known to propagate along penetrating arterioles to pial arterioles, but it is unclear whether spontaneous vasomotion propagates similarly. We therefore imaged both stimulus-evoked and spontaneous changes in pial arteriole diameter in awake, head-fixed mice with 2-photon microscopy. By cross-correlating different regions of interest (ROIs) along the length of imaged arterioles, we assessed vasomotion propagation. We found that both during rest and during visual stimulation, one-third of the arterioles showed significant propagation (i.e., a wave), with a median (interquartile range) wave speed of 405 (323) µm/s at rest and 345 (177) µm/s during stimulation. In a second group of mice, with GCaMP expression in their vascular smooth muscle cells, we also found spontaneous propagation of calcium signaling along pial arterioles. In summary, we demonstrate that spontaneous vasomotion propagates along pial arterioles like stimulus-evoked vascular reactivity.

Blood-brain barrier leakage and perivascular inflammation in cerebral amyloid angiopathy.

Cerebral amyloid angiopathy is a small vessel disease associated with cortical microbleeds and lobar intracerebral haemorrhage due to amyloid-ß deposition in the walls of leptomeningeal and cortical arterioles. The mechanisms of cerebral amyloid angiopathy-related haemorrhage remain largely unknown. Recent work has demonstrated that ruptured blood vessels have limited (or no) amyloid-ß at the site of bleeding and evidence of local vascular remodelling. We hypothesized that blood-brain barrier leakage and perivascular inflammation may be involved in this remodelling process. This study examined cortical arterioles at various stages of cerebral amyloid angiopathy-related vascular pathology (without evidence of microhaemorrhage) in autopsy tissue from seven cases with definite cerebral amyloid angiopathy. We included temporo-occipital sections with microbleeds guided by ex vivo MRI from two cases with severe cerebral amyloid angiopathy and systematically sampled occipital sections from five consecutive cases with varying cerebral amyloid angiopathy severity. Haematoxylin and eosin stains and immunohistochemistry against amyloid-ß, fibrin(ogen), smooth muscle actin, reactive astrocytes (glial fibrillary acidic protein) and activated microglia (cluster of differentiation 68) were performed. Arterioles were graded using a previously proposed scale of individual vessel cerebral amyloid angiopathy severity, and a blinded assessment for blood-brain barrier leakage, smooth muscle actin and perivascular inflammation was performed. Blood-brain barrier leakage and smooth muscle actin loss were observed in significantly more vessels with mild amyloid-ß deposition (Grade 1 vessels; P = 0.044 and P = 0.012, respectively) as compared to vessels with no amyloid-ß (Grade 0), and blood-brain barrier leakage was observed in 100% of vessels with evidence of vessel remodelling (Grades 3 and 4). Perivascular inflammation in the form of reactive astrocytes and activated microglia was observed predominantly surrounding arterioles at later stages of vessel pathology (Grades 2-4) and consistently around vessels with the same morphological features as ruptured vessel segments (Grade 4). These findings suggest a role for blood-brain barrier leakage and perivascular inflammation leading to arteriolar remodelling and haemorrhage in cerebral amyloid angiopathy, with early blood-brain barrier leakage as a potential trigger for subsequent perivascular inflammation.

Corpus callosum lesions are associated with worse cognitive performance in cerebral amyloid angiopathy.

The impact of vascular lesions on cognition is location dependent. Here, we assessed the contribution of small vessel disease lesions in the corpus callosum to vascular cognitive impairment in cerebral amyloid angiopathy, as a model for cerebral small vessel disease. Sixty-five patients with probable cerebral amyloid angiopathy underwent 3T magnetic resonance imaging, including a diffusion tensor imaging scan, and neuropsychological testing. Microstructural white-matter integrity was quantified by fractional anisotropy and mean diffusivity. Z-scores on individual neuropsychological tests were averaged into five cognitive domains: information processing speed, executive functioning, memory, language and visuospatial ability. Corpus callosum lesions were defined as haemorrhagic (microbleeds or larger bleeds) or ischaemic (microinfarcts, larger infarcts and diffuse fluid-attenuated inversion recovery hyperintensities). Associations between corpus callosum lesion presence, microstructural white-matter integrity and cognitive performance were examined with multiple regression models. The prevalence of corpus callosum lesions was confirmed in an independent cohort of memory clinic patients with and without cerebral amyloid angiopathy (n = 82). In parallel, we assessed corpus callosum lesions on ex vivo magnetic resonance imaging in cerebral amyloid angiopathy patients (n = 19) and controls (n = 5) and determined associated tissue abnormalities with histopathology. A total number of 21 corpus callosum lesions was found in 19/65 (29%) cerebral amyloid angiopathy patients. Corpus callosum lesion presence was associated with reduced microstructural white-matter integrity within the corpus callosum and in the whole-brain white matter. Patients with corpus callosum lesions performed significantly worse on all cognitive domains except language, compared with those without corpus callosum lesions after correcting for age, sex, education and time between magnetic resonance imaging and neuropsychological assessment. This association was independent of the presence of intracerebral haemorrhage, whole-brain fractional anisotropy and mean diffusivity, and white-matter hyperintensity volume and brain volume for the domains of information processing speed and executive functioning. In the memory clinic patient cohort, corpus callosum lesions were present in 14/54 (26%) patients with probable and 2/8 (25%) patients with possible cerebral amyloid angiopathy, and in 3/20 (15%) patients without cerebral amyloid angiopathy. In the ex vivo cohort, corpus callosum lesions were present in 10/19 (53%) patients and 2/5 (40%) controls. On histopathology, ischaemic corpus callosum lesions were associated with tissue loss and demyelination, which extended beyond the lesion core. Together, these data suggest that corpus callosum lesions are a frequent finding in cerebral amyloid angiopathy, and that they independently contribute to cognitive impairment through strategic microstructural disruption of white-matter tracts.

Perivascular space dilation is associated with vascular amyloid-ß 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-ß (Aß) 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 Aß 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.

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

Traditionally, analysis of neuropathological markers in neurodegenerative diseases has relied on visual assessments of stained sections. Resulting semiquantitative scores often vary between individual raters and research centers, limiting statistical approaches. To overcome these issues, we have developed six deep learning-based models, that identify some of the most characteristic markers of Alzheimer's disease (AD) and cerebral amyloid angiopathy (CAA). The deep learning-based models are trained to differentially detect parenchymal amyloid ß (Aß)-plaques, vascular Aß-deposition, iron and calcium deposition, reactive astrocytes, microglia, as well as fibrin extravasation. The models were trained on digitized histopathological slides from brains of patients with AD and CAA, using a workflow that allows neuropathology experts to train convolutional neural networks (CNNs) on a cloud-based graphical interface. Validation of all models indicated a very good to excellent performance compared to three independent expert human raters. Furthermore, the Aß and iron models were consistent with previously acquired semiquantitative scores in the same dataset and allowed the use of more complex statistical approaches. For example, linear mixed effects models could be used to confirm the previously described relationship between leptomeningeal CAA severity and cortical iron accumulation. A similar approach enabled us to explore the association between neuroinflammation and disparate Aß pathologies. The presented workflow is easy for researchers with pathological expertise to implement and is customizable for additional histopathological markers. The implementation of deep learning-assisted analyses of histopathological slides is likely to promote standardization of the assessment of neuropathological markers across research centers, which will allow specific pathophysiological questions in neurodegenerative disease to be addressed in a harmonized way and on a larger scale.