AI-based differential diagnosis of dementia etiologies on multimodal data.

Differential diagnosis of dementia remains a challenge in neurology due to symptom overlap across etiologies, yet it is crucial for formulating early, personalized management strategies. Here, we present an AI model that harnesses a broad array of data, including demographics, individual and family medical history, medication use, neuropsychological assessments, functional evaluations, and multimodal neuroimaging, to identify the etiologies contributing to dementia in individuals. The study, drawing on 51,269 participants across 9 independent, geographically diverse datasets, facilitated the identification of 10 distinct dementia etiologies. It aligns diagnoses with similar management strategies, ensuring robust predictions even with incomplete data. Our model achieved a micro-averaged area under the receiver operating characteristic curve (AUROC) of 0.94 in classifying individuals with normal cognition, mild cognitive impairment and dementia. Also, the micro-averaged AUROC was 0.96 in differentiating the dementia etiologies. Our model demonstrated proficiency in addressing mixed dementia cases, with a mean AUROC of 0.78 for two co-occurring pathologies. In a randomly selected subset of 100 cases, the AUROC of neurologist assessments augmented by our AI model exceeded neurologist-only evaluations by 26.25%. Furthermore, our model predictions aligned with biomarker evidence and its associations with different proteinopathies were substantiated through postmortem findings. Our framework has the potential to be integrated as a screening tool for dementia in various clinical settings and drug trials, with promising implications for person-level management.

Delivering a Diagnosis of Parkinson's Disease and Parkinsonism with Wisdom and Sensitivity.

Parkinson's disease (PD) is a chronic, progressive, complex movement disorder. In addition to the motor manifestations, changes in mood and cognition frequently occur. It is understandable that receiving this diagnosis can be difficult for patients and their significant others. For the clinician, delivering a PD diagnosis can be challenging and requires a comprehensive patient assessment followed by a thoughtful treatment plan. How this diagnosis is conveyed can have a long-term impact on patient outcomes such as treatment adherence, participation in decision making, understanding of PD, and satisfaction with care. Because a PD diagnosis is often complicated by uncertainty about the diagnosis itself as well as future prognosis, a sensitive patient-centered approach to care, balanced with realistic expectations, is recommended. Full disclosure, honesty, and empathy on the part of the entire healthcare team are required. This includes relevant information tailored to the patient's unique needs at the time of diagnosis as well as referrals to appropriate rehabilitation and support services. Consistent, timely follow-up of all interventions is essential. It is essential that a diagnosis of PD is properly delivered to optimize understanding of PD, treatment adherence, participation in decision making, and satisfaction with care. In this article, we provide guidance on delivery of this diagnosis based on a growing body of evidence and our >35-year collective clinical experience and work developing and utilizing pertinent, creative educational tools and comprehensive, sensitive support programs for newly diagnosed patients and their significant others. Although most of the evidence we present pertains to PD, our experience suggests it could also apply to other forms of Parkinsonism and other chronic or progressive movement disorders.

Multimodal deep learning for Alzheimer's disease dementia assessment.

Worldwide, there are nearly 10 million new cases of dementia annually, of which Alzheimer's disease (AD) is the most common. New measures are needed to improve the diagnosis of individuals with cognitive impairment due to various etiologies. Here, we report a deep learning framework that accomplishes multiple diagnostic steps in successive fashion to identify persons with normal cognition (NC), mild cognitive impairment (MCI), AD, and non-AD dementias (nADD). We demonstrate a range of models capable of accepting flexible combinations of routinely collected clinical information, including demographics, medical history, neuropsychological testing, neuroimaging, and functional assessments. We then show that these frameworks compare favorably with the diagnostic accuracy of practicing neurologists and neuroradiologists. Lastly, we apply interpretability methods in computer vision to show that disease-specific patterns detected by our models track distinct patterns of degenerative changes throughout the brain and correspond closely with the presence of neuropathological lesions on autopsy. Our work demonstrates methodologies for validating computational predictions with established standards of medical diagnosis.

Development and validation of an interpretable deep learning framework for Alzheimer's disease classification.

Alzheimer's disease is the primary cause of dementia worldwide, with an increasing morbidity burden that may outstrip diagnosis and management capacity as the population ages. Current methods integrate patient history, neuropsychological testing and MRI to identify likely cases, yet effective practices remain variably applied and lacking in sensitivity and specificity. Here we report an interpretable deep learning strategy that delineates unique Alzheimer's disease signatures from multimodal inputs of MRI, age, gender, and Mini-Mental State Examination score. Our framework linked a fully convolutional network, which constructs high resolution maps of disease probability from local brain structure to a multilayer perceptron and generates precise, intuitive visualization of individual Alzheimer's disease risk en route to accurate diagnosis. The model was trained using clinically diagnosed Alzheimer's disease and cognitively normal subjects from the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset (n = 417) and validated on three independent cohorts: the Australian Imaging, Biomarker and Lifestyle Flagship Study of Ageing (AIBL) (n = 382), the Framingham Heart Study (n = 102), and the National Alzheimer's Coordinating Center (NACC) (n = 582). Performance of the model that used the multimodal inputs was consistent across datasets, with mean area under curve values of 0.996, 0.974, 0.876 and 0.954 for the ADNI study, AIBL, Framingham Heart Study and NACC datasets, respectively. Moreover, our approach exceeded the diagnostic performance of a multi-institutional team of practicing neurologists (n = 11), and high-risk cerebral regions predicted by the model closely tracked post-mortem histopathological findings. This framework provides a clinically adaptable strategy for using routinely available imaging techniques such as MRI to generate nuanced neuroimaging signatures for Alzheimer's disease diagnosis, as well as a generalizable approach for linking deep learning to pathophysiological processes in human disease.

Cerebrospinal fluid, plasma, and saliva in the BioFIND study: Relationships among biomarkers and Parkinson's disease Features.

OBJECTIVE: Examine relationships among neurodegenerative biomarkers and PD motor and nonmotor symptoms. BACKGROUND: CSF alpha-synuclein is decreased in PD versus healthy controls, but whether plasma and saliva alpha-synuclein differentiate these groups is controversial. Correlations of alpha-synuclein among biofluids (CSF, plasma, saliva) or biomarkers (eg, beta-amyloid, tau [total, phosphorylated]) are not fully understood. The relationships of these biomarkers with PD clinical features remain unclear. METHODS: BioFIND, a cross-sectional, observational study, examines clinical and biomarker characteristics in moderate-advanced PD and matched healthy controls. We compared alpha-synuclein concentrations across diagnosis, biofluids, and CSF biomarkers. Correlations of CSF biomarkers and MDS-UPDRS, motor phenotype, MoCA, and rapid eye movement sleep behavior disorder questionnaire scores in PD were examined. RESULTS: CSF alpha-synuclein was lower in PD versus controls (P = .01), controlling for age, gender, and education. Plasma and saliva alpha-synuclein did not differ between PD and controls, and alpha-synuclein did not significantly correlate among biofluids. CSF beta-amyloid1-42 was lower in PD versus controls (P < .01), and correlated weakly with MoCA recall scores (r = 0.23, P = .02). CSF alpha-synuclein was lower in the postural instability/gait difficulty phenotype than other motor phenotypes (P < .01). No CSF biomarkers predicted or correlated with total motor or rapid eye movement sleep behavior disorder scores. CSF alpha-synuclein correlated with beta-amyloid1-42 , total-tau, and phosphorylated-tau (r = 0.41, 0.81, 0.43, respectively; Ps < .001). CONCLUSION: Lower CSF alpha-synuclein is associated with diagnosis and motor phenotype in moderate-advanced PD. Plasma and saliva alpha-synuclein neither correlate with CSF alpha-synuclein, nor distinguish PD from controls. CSF beta-amyloid1-42 remains a potential biomarker for cognitive impairment in PD. © 2017 The Authors. Movement Disorders published by Wiley Periodicals, Inc. on behalf of International Parkinson and Movement Disorder Society.

A randomized trial of inhaled levodopa (CVT-301) for motor fluctuations in Parkinson's disease.

BACKGROUND: Although levodopa is the most effective oral PD therapy, many patients experience motor fluctuations, including sudden loss of dose effect and delayed benefit. CVT-301 is a levodopa inhalation powder with the potential for rapid onset of action. The objective of this study was to evaluate CVT-301 self-administered by PD patients to relieve OFF episodes. METHODS: PD patients with ≥2 hours per day of OFF time despite oral levodopa ≥4 times per day were randomized to CVT-301 or placebo for 4 weeks, to be used up to 3 times per day for OFF episodes. After 2 weeks, the study-drug dose was escalated from 35 to 50 mg. The primary end point was mean change in UPDRS Part III score from a predose OFF state to the average of postdose scores obtained at 10, 20, 30, and 60 minutes, as assessed in-clinic at the end of week 4. Home diaries were recorded. RESULTS: Eighty-six patients used the study drug at an average frequency of 2.1 times per day for CVT-301 and for placebo. At 4 weeks, least-squares mean change in UPDRS Part III score favored CVT-301 by 7.0 points (P < 0.001). A treatment effect was evident at 10 minutes. At 4 weeks, least-squares mean OFF-time change from baseline favored CVT-301 by 0.9 hours per day (P = 0.045). The most frequently reported adverse events in the CVT-301 group were dizziness, cough, and nausea, each in 7% (3 of 43 patients). CONCLUSIONS: CVT-301 self-administered during OFF episodes provided rapid improvement of motor function, and daily OFF time was significantly reduced at the higher dose. CVT-301 was generally safe and well-tolerated. © 2016 The Authors. Movement Disorders published by Wiley Periodicals, Inc. on behalf of International Parkinson and Movement Disorder Society.

Head injury at early ages is associated with risk of Parkinson's disease.

INTRODUCTION: The literature on the effect of head injuries on the risk of PD is inconclusive. Some researchers have hypothesized that studies that have seen an effect are simply capturing injury related to pre-clinical PD. However in animal models brain inflammation, which can be initiated by head trauma, has been shown to produce PD-like effects. Furthermore, animal studies have found that early life inflammation in particular is of relevance for PD pathology. METHODS: We conducted an unmatched case-control study of 379 neurologist confirmed PD patients and 230 controls from the greater Boston, Massachusetts area with questionnaire data on history of head injury and other covariates. We used multivariable logistic regression to estimate adjusted odds ratios (OR) and their corresponding 95% confidence intervals (CI) for PD. RESULTS: When we excluded injuries that occurred less than 10 years prior to the diagnosis of PD (in order to avoid reverse causation), we found an increased risk of PD associated with a head injury that resulted in a loss of consciousness, but it did not reach statistical significance (OR = 1.57; 95% CI = 0.89-2.80). We found a significant (p = 0.04) effect of age at first head injury. For every 5 year earlier age at first head injury with loss of consciousness the OR for PD was 1.37 (95% CI: 1.01-1.86). CONCLUSION: Our results suggest that head injury in early life increases the risk of PD.

Candidate glutamatergic and dopaminergic pathway gene variants do not influence Huntington's disease motor onset.

Huntington's disease (HD) is a neurodegenerative disorder characterized by motor, cognitive, and behavioral disturbances. It is caused by the expansion of the HTT CAG repeat, which is the major determinant of age at onset (AO) of motor symptoms. Aberrant function of N-methyl-D-aspartate receptors and/or overexposure to dopamine has been suggested to cause significant neurotoxicity, contributing to HD pathogenesis. We used genetic association analysis in 1,628 HD patients to evaluate candidate polymorphisms in N-methyl-D-aspartate receptor subtype genes (GRIN2A rs4998386 and rs2650427, and GRIN2B rs1806201) and functional polymorphisms in genes in the dopamine pathway (DAT1 3' UTR 40-bp variable number tandem repeat (VNTR), DRD4 exon 3 48-bp VNTR, DRD2 rs1800497, and COMT rs4608) as potential modifiers of the disease process. None of the seven polymorphisms tested was found to be associated with significant modification of motor AO, either in a dominant or additive model, after adjusting for ancestry. The results of this candidate-genetic study therefore do not provide strong evidence to support a modulatory role for these variations within glutamatergic and dopaminergic genes in the AO of HD motor manifestations.

TAA repeat variation in the GRIK2 gene does not influence age at onset in Huntington's disease.

Huntington's disease is a neurodegenerative disorder caused by an expanded CAG trinucleotide repeat whose length is the major determinant of age at onset but remaining variation appears to be due in part to the effect of genetic modifiers. GRIK2, which encodes GluR6, a mediator of excitatory neurotransmission in the brain, has been suggested in several studies to be a modifier gene based upon a 3' untranslated region TAA trinucleotide repeat polymorphism. Prior to investing in detailed studies of the functional impact of this polymorphism, we sought to confirm its effect on age at onset in a much larger dataset than in previous investigations. We genotyped the HD CAG repeat and the GRIK2 TAA repeat in DNA samples from 2,911 Huntington's disease subjects with known age at onset, and tested for a potential modifier effect of GRIK2 using a variety of statistical approaches. Unlike previous reports, we detected no evidence of an influence of the GRIK2 TAA repeat polymorphism on age at motor onset. Similarly, the GRIK2 polymorphism did not show significant modifier effect on psychiatric and cognitive age at onset in HD. Comprehensive analytical methods applied to a much larger sample than in previous studies do not support a role for GRIK2 as a genetic modifier of age at onset of clinical symptoms in Huntington's disease.

Population stratification may bias analysis of PGC-1α as a modifier of age at Huntington disease motor onset.

Huntington's disease (HD) is an inherited neurodegenerative disorder characterized by motor, cognitive and behavioral disturbances, caused by the expansion of a CAG trinucleotide repeat in the HD gene. The CAG allele size is the major determinant of age at onset (AO) of motor symptoms, although the remaining variance in AO is highly heritable. The rs7665116 SNP in PPARGC1A, encoding the mitochondrial regulator PGC-1α, has been reported to be a significant modifier of AO in three European HD cohorts, perhaps due to affected cases from Italy. We attempted to replicate these findings in a large collection of (1,727) HD patient DNA samples of European origin. In the entire cohort, rs7665116 showed a significant effect in the dominant model (p value = 0.008) and the additive model (p value = 0.009). However, when examined by origin, cases of Southern European origin had an increased rs7665116 minor allele frequency (MAF), consistent with this being an ancestry-tagging SNP. The Southern European cases, despite similar mean CAG allele size, had a significantly older mean AO (p < 0.001), suggesting population-dependent phenotype stratification. When the generalized estimating equations models were adjusted for ancestry, the effect of the rs7665116 genotype on AO decreased dramatically. Our results do not support rs7665116 as a modifier of AO of motor symptoms, as we found evidence for a dramatic effect of phenotypic (AO) and genotypic (MAF) stratification among European cohorts that was not considered in previously reported association studies. A significantly older AO in Southern Europe may reflect population differences in genetic or environmental factors that warrant further investigation.

Common SNP-based haplotype analysis of the 4p16.3 Huntington disease gene region.

Age at the onset of motor symptoms in Huntington disease (HD) is determined largely by the length of a CAG repeat expansion in HTT but is also influenced by other genetic factors. We tested whether common genetic variation near the mutation site is associated with differences in the distribution of expanded CAG alleles or age at the onset of motor symptoms. To define disease-associated single-nucleotide polymorphisms (SNPs), we compared 4p16.3 SNPs in HD subjects with population controls in a case:control strategy, which revealed that the strongest signals occurred at a great distance from the HD mutation as a result of "synthetic association" with SNP alleles that are of low frequency in population controls. Detailed analysis delineated a prominent ancestral haplotype that accounted for ∼50% of HD chromosomes and extended to at least 938 kb on about half of these. Together, the seven most abundant haplotypes accounted for ∼83% of HD chromosomes. Neither the extended shared haplotype nor the individual local HTT haplotypes were associated with altered CAG-repeat length distribution or residual age at the onset of motor symptoms, arguing against modification of these disease features by common cis-regulatory elements. Similarly, the 11 most frequent control haplotypes showed no trans-modifier effect on age at the onset of motor symptoms. Our results argue against common local regulatory variation as a factor influencing HD pathogenesis, suggesting that genetic modifiers be sought elsewhere in the genome. They also indicate that genome-wide association analysis with a small number of cases can be effective for regional localization of genetic defects, even when a founder effect accounts for only a fraction of the disorder.

Association of cumulative lead exposure with Parkinson's disease.

BACKGROUND: Research using reconstructed exposure histories has suggested an association between heavy metal exposures, including lead, and Parkinson's disease (PD), but the only study that used bone lead, a biomarker of cumulative lead exposure, found a nonsignificant increase in risk of PD with increasing bone lead. OBJECTIVES: We sought to assess the association between bone lead and PD. METHODS: Bone lead concentrations were measured using 109Cd excited K-shell X-ray fluorescence from 330 PD patients (216 men, 114 women) and 308 controls (172 men, 136 women) recruited from four clinics for movement disorders and general-community cohorts. Adjusted odds ratios (ORs) for PD were calculated using logistic regression. RESULTS: The average age of cases and controls at bone lead measurement was 67 (SD = 10) and 69 (SD = 9) years of age, respectively. In primary analyses of cases and controls recruited from the same groups, compared with the lowest quartile of tibia lead, the OR for PD in the highest quartile was 3.21 [95% confidence interval (CI), 1.17-8.83]. Results were similar but slightly weaker in analyses restricted to cases and controls recruited from the movement disorders clinics only (fourth-quartile OR = 2.57; 95% CI, 1.11-5.93) or when we included controls recruited from sites that did not also contribute cases (fourth-quartile OR = 1.91; 95% CI, 1.01-3.60). We found no association with patella bone lead. CONCLUSIONS: These findings, using an objective biological marker of cumulative lead exposure among typical PD patients seen in our movement disorders clinics, strengthen the evidence that cumulative exposure to lead increases the risk of PD.