Evaluating the performance of large language models: ChatGPT and Google Bard in generating differential diagnoses in clinicopathological conferences of neurodegenerative disorders.

This study explores the utility of the large language models (LLMs), specifically ChatGPT and Google Bard, in predicting neuropathologic diagnoses from clinical summaries. A total of 25 cases of neurodegenerative disorders presented at Mayo Clinic brain bank Clinico-Pathological Conferences were analyzed. The LLMs provided multiple pathologic diagnoses and their rationales, which were compared with the final clinical diagnoses made by physicians. ChatGPT-3.5, ChatGPT-4, and Google Bard correctly made primary diagnoses in 32%, 52%, and 40% of cases, respectively, while correct diagnoses were included in 76%, 84%, and 76% of cases, respectively. These findings highlight the potential of artificial intelligence tools like ChatGPT in neuropathology, suggesting they may facilitate more comprehensive discussions in clinicopathological conferences.

Validation Study of the MDS Criteria for the Diagnosis of Multiple System Atrophy in the Mayo Clinic Brain Bank.

BACKGROUND AND OBJECTIVE: The second consensus criteria in 2008 have been used in diagnosing multiple system atrophy (MSA). The International Parkinson and Movement Disorder Society (MDS) proposed new diagnostic criteria for MSA in 2022. This study aimed to compare the diagnostic accuracy between these 2 criteria and validate the clinical utility of the newly proposed criteria for MSA. METHODS: We conducted a retrospective autopsy cohort study of consecutive patients with a clinical or pathologic diagnosis of MSA from the Mayo Clinic brain bank between 1998 and 2021. We studied 352 patients (250 pathologically diagnosed MSA and 102 non-MSA); MDS criteria and the second consensus criteria were applied. The sensitivity, specificity, and area under the curve (AUC) of receiver operating characteristic curves were compared between these criteria. Comparison was conducted between clinical subtypes and among clinically challenging cases (those with different clinical diagnoses or those with suspected but undiagnosed MSA before death). We also used machine learning algorithm, eXtreme Gradient Boosting, to identify clinical features contributing diagnostic performance. RESULTS: The sensitivity and specificity of clinically established and probable MSA by the MDS criteria were 16% and 99% and 64% and 74%, respectively. The sensitivity and specificity of probable MSA and possible MSA by the second consensus criteria were 72% and 52% and 93% and 21%, respectively. The AUC of MDS clinically probable MSA was the highest (0.69). The diagnostic performance did not differ between clinical subtypes. In clinically challenging cases, MDS clinically established MSA maintained high specificity and MDS clinically probable MSA demonstrated the highest AUC (0.62). MRI findings contributed to high specificity. In addition, combining core clinical features with 2 or more from any of the 13 supporting features and the absence of exclusion criteria also yielded high specificity. Among supporting features, rapid progression was most important for predicting MSA pathology. DISCUSSION: The MDS criteria showed high specificity with clinically established MSA and moderate sensitivity and specificity with clinically probable MSA. The observation that high specificity could be achieved with clinical features alone suggests that MSA diagnosis with high specificity is possible even in areas where MRI is not readily available.

Brain Bank Questionnaire Helps in Differential Diagnosis of Movement Disorders: An Autopsy Study of 150 Patients.

Background: As part of the CurePSP brain donation program, a questionnaire was developed to gather basic clinical information on donors; however, its usefulness has not been evaluated. Objective: To assess the value of information obtained from the questionnaire in differentiating between parkinsonian disorders. Methods: We reviewed 150 questionnaires, including 50 patients, each with a neuropathologic diagnosis of Lewy body disease (LBD), multiple system atrophy (MSA), or progressive supranuclear palsy. The frequency of clinical features recorded in the questionnaires was compared for the three disorders, and a machine learning algorithm was used to identify features predicting neuropathologic diagnosis. Results: The information from the questionnaires correlated with core clinical features for each disorder, such as hallucinations for LBD and autonomic dysfunction for MSA. Hallucinations and disorientations were identified as the key variables that contributed most to the prediction of neuropathology. Conclusion: The questionnaire provides useful clinical information for clinicopathological correlative studies.

Lysosomal polygenic risk is associated with the severity of neuropathology in Lewy body disease.

Intraneuronal accumulation of misfolded α-synuclein is the pathological hallmark of Parkinson's disease and dementia with Lewy bodies, often co-occurring with variable degrees of Alzheimer's disease related neuropathology. Genetic association studies have successfully identified common variants associated with disease risk and phenotypic traits in Lewy body disease, yet little is known about the genetic contribution to neuropathological heterogeneity. Using summary statistics from Parkinson's disease and Alzheimer's disease genome-wide association studies, we calculated polygenic risk scores and investigated the relationship with Lewy, amyloid-ß and tau pathology. Associations were nominated in neuropathologically defined samples with Lewy body disease from the Netherlands Brain Bank (n = 217) and followed up in an independent sample series from the Mayo Clinic Brain Bank (n = 394). We also generated stratified polygenic risk scores based on single-nucleotide polymorphisms annotated to eight functional pathways or cell types previously implicated in Parkinson's disease and assessed for association with Lewy pathology in subgroups with and without significant Alzheimer's disease co-pathology. In an ordinal logistic regression model, the Alzheimer's disease polygenic risk score was associated with concomitant amyloid-ß and tau pathology in both cohorts. Moreover, both cohorts showed a significant association between lysosomal pathway polygenic risk and Lewy pathology, which was more consistent than the association with a general Parkinson's disease risk score and specific to the subset of samples without significant concomitant Alzheimer's disease related neuropathology. Our findings provide proof of principle that the specific risk alleles a patient carries for Parkinson's and Alzheimer's disease also influence key aspects of the underlying neuropathology in Lewy body disease. The interrelations between genetic architecture and neuropathology are complex, as our results implicate lysosomal risk loci specifically in the subset of samples without Alzheimer's disease co-pathology. Our findings hold promise that genetic profiling may help predict the vulnerability to specific neuropathologies in Lewy body disease, with potential relevance for the further development of precision medicine in these disorders.

Creutzfeldt-Jakob Disease Misdiagnosed as Multiple System Atrophy.

Background: Multiple system atrophy (MSA) is a progressive neurodegenerative disorder characterized by various combinations of autonomic failure, parkinsonism, and cerebellar syndromes. Although consensus criteria have been widely used to diagnose MSA, accurate clinical diagnosis remains challenging. Other neurodegenerative disorders, such as Lewy body disease, can mimic MSA. Objectives: We described clinical and neuropathologic findings of two patients with Creutzfeldt-Jakob disease (CJD) who had antemortem clinical diagnoses of MSA. Methods: The brain bank for neurodegenerative disorders was queried for cases with a clinical diagnosis of MSA, but neuropathologic findings of CJD. Results: Case 1 was a 55-year-old man with a 6-month history of orthostatic hypotension, parkinsonism, cerebellar ataxia, bradyphrenia, and memory impairment. Case 2 was a 65-year-old man who had a 5-year history of cerebellar ataxia, parkinsonism, and cognitive impairment, as well as a 7-year history of dream enactment behavior. Neither case had characteristic α-synuclein immunoreactive neuronal or glial inclusions typical of MSA. Instead, they had spongiform encephalopathy with neuronal loss and gliosis with prion protein-immunoreactive kuru-like plaques. Genetic analyses in case 1 had wild-type PRNP, whereas case 2 revealed a 4-octapeptide repeat insertion in PRNP. Conclusions: Even when clinical features suggest MSA, CJD should also be considered if the progression is rapid or the disease course is atypical, such as the absence of autonomic dysfunction for an extended period.

Diagnosis of Alzheimer Disease and Tauopathies on Whole-Slide Histopathology Images Using a Weakly Supervised Deep Learning Algorithm.

Neuropathologic assessment during autopsy is the gold standard for diagnosing neurodegenerative disorders. Neurodegenerative conditions, such as Alzheimer disease (AD) neuropathological change, are a continuous process from normal aging rather than categorical; therefore, diagnosing neurodegenerative disorders is a complicated task. We aimed to develop a pipeline for diagnosing AD and other tauopathies, including corticobasal degeneration (CBD), globular glial tauopathy, Pick disease, and progressive supranuclear palsy. We used a weakly supervised deep learning-based approach called clustering-constrained-attention multiple-instance learning (CLAM) on the whole-slide images (WSIs) of patients with AD (n = 30), CBD (n = 20), globular glial tauopathy (n = 10), Pick disease (n = 20), and progressive supranuclear palsy (n = 20), as well as nontauopathy controls (n = 21). Three sections (A: motor cortex; B: cingulate gyrus and superior frontal gyrus; and C: corpus striatum) that had been immunostained for phosphorylated tau were scanned and converted to WSIs. We evaluated 3 models (classic multiple-instance learning, single-attention-branch CLAM, and multiattention-branch CLAM) using 5-fold cross-validation. Attention-based interpretation analysis was performed to identify the morphologic features contributing to the classification. Within highly attended regions, we also augmented gradient-weighted class activation mapping to the model to visualize cellular-level evidence of the model's decisions. The multiattention-branch CLAM model using section B achieved the highest area under the curve (0.970 ± 0.037) and diagnostic accuracy (0.873 ± 0.087). A heatmap showed the highest attention in the gray matter of the superior frontal gyrus in patients with AD and the white matter of the cingulate gyrus in patients with CBD. Gradient-weighted class activation mapping showed the highest attention in characteristic tau lesions for each disease (eg, numerous tau-positive threads in the white matter inclusions for CBD). Our findings support the feasibility of deep learning-based approaches for the classification of neurodegenerative disorders on WSIs. Further investigation of this method, focusing on clinicopathologic correlations, is warranted.

APOE2 Exacerbates TDP-43 Related Toxicity in the Absence of Alzheimer Pathology.

OBJECTIVE: Recent evidence supports a link between increased TDP-43 burden and the presence of an APOE4 gene allele in Alzheimer's disease (AD); however, it is difficult to conclude the direct effect of APOE on TDP-43 pathology due to the presence of mixed AD pathologies. The goal of this study is to address how APOE isoforms impact TDP-43 pathology and related neurodegeneration in the absence of typical AD pathologies. METHODS: We overexpressed human TDP-43 via viral transduction in humanized APOE2, APOE3, APOE4 mice, and murine Apoe-knockout (Apoe-KO) mice. Behavior tests were performed across ages. Animals were harvested at 11 months of age and TDP-43 overexpression-related neurodegeneration and gliosis were assessed. To further address the human relevance, we analyzed the association of APOE with TDP-43 pathology in 160 postmortem brains from autopsy-confirmed amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration with motor neuron disease (FTLD-MND) in the Mayo Clinic Brain Bank. RESULTS: We found that TDP-43 overexpression induced motor function deficits, neuronal loss, and gliosis in the motor cortex, especially in APOE2 mice, with much milder or absent effects in APOE3, APOE4, or Apoe-KO mice. In the motor cortex of the ALS and FTLD-MND postmortem human brains, we found that the APOE2 allele was associated with more severe TDP-43-positive dystrophic neurites. INTERPRETATION: Our data suggest a genotype-specific effect of APOE on TDP-43 proteinopathy and neurodegeneration in the absence of AD pathology, with the strongest association seen with APOE2. ANN NEUROL 2023;93:830-843.

Galactosylceramidase deficiency and pathological abnormalities in cerebral white matter of Krabbe disease.

Krabbe Disease (KD) is an autosomal recessive disorder that results from loss-of-function mutations in the GALC gene, which encodes lysosomal enzyme galactosylceramidase (GALC). Functional deficiency of GALC is toxic to myelin-producing cells, which leads to progressive demyelination in both the central and peripheral nervous systems. It is hypothesized that accumulation of psychosine, which can only be degraded by GALC, is a primary initiator of pathologic cascades. Despite the central role of GALC in KD pathomechanism, investigations of GALC deficiency at a protein level are largely absent, due in part, to the lack of sensitive antibodies in the field. Leveraging two custom antibodies that can detect GALC at endogenous levels, we demonstrated that GALC protein is predominantly localized to oligodendrocytes in cerebral white matter of an infant brain, consistent with its functional role in myelination. Mature GALC could also be quantitatively detected as a 26 kDa band by western blotting and correlated to enzyme activity in brain tissues. The p.Ile562Thr polymorphic variant, which is over-represented in the KD population, was associated with reduced mature GALC protein and activity. In three infantile KD cases, homozygous null mutations in GALC lead to deficiency in total GALC protein and activity. Interestingly, although GALC activity was absent, normal levels of total GALC protein were detected by a sandwich ELISA using our custom antibodies in a later-onset KD brain, which suggests that the assay has the potential to differentiate infantile- and later-onset KD cases. Among the infantile KD cases, we quantified a 5-fold increase in psychosine levels, and observed increased levels of acid ceramidase, a key enzyme for psychosine production, and hyperglycosylated lysosomal-associated membrane protein 1, a marker for lysosomal activation, in periventricular white matter, a major pathological brain region, when compared with age-matched normal controls. While near complete demyelination was observed in these cases, we quantified that an early-infantile case (age of death at 10 months) had about 3-fold increases in both globoid cells, a pathological hallmark for KD, and CD8-positive T lymphocytes, a pathological marker for multiple sclerosis, in the white matter when compared with a slower progressing infantile case (age of death at 21 months), which suggests a positive correlation between clinical severity and neuropathology. Taken together, our findings have advanced the understanding of GALC protein biology in the context of normal and KD brain white matter. We also revealed new neuropathological changes that may provide insights to understand KD pathogenesis.

AAV liver expression of FIX-Padua prevents and eradicates FIX inhibitor without increasing thrombogenicity in hemophilia B dogs and mice.

Emerging successful clinical data on gene therapy using adeno-associated viral (AAV) vector for hemophilia B (HB) showed that the risk of cellular immune response to vector capsid is clearly dose dependent. To decrease the vector dose, we explored AAV-8 (1-3 × 10(12) vg/kg) encoding a hyperfunctional factor IX (FIX-Padua, arginine 338 to leucine) in FIX inhibitor-prone HB dogs. Two naïve HB dogs showed sustained expression of FIX-Padua with an 8- to 12-fold increased specific activity reaching 25% to 40% activity without antibody formation to FIX. A third dog with preexisting FIX inhibitors exhibited a transient anamnestic response (5 Bethesda units) at 2 weeks after vector delivery following by spontaneous eradication of the antibody to FIX by day 70. In this dog, sustained FIX expression reached ∼200% and 30% of activity and antigen levels, respectively. Immune tolerance was confirmed in all dogs after challenges with plasma-derived FIX concentrate. Shortening of the clotting times and lack of bleeding episodes support the phenotypic correction of the severe phenotype, with no clinical or laboratory evidence of risk of thrombosis. Provocative studies in mice showed that FIX-Padua exhibits similar immunogenicity and thrombogenicity compared with FIX wild type. Collectively, these data support the potential translation of gene-based strategies using FIX-Padua for HB.