Josephs, Keith A.

The objective of this study was to characterize network-level changes in nonfluent/agrammatic Primary Progressive Aphasia (agPPA) and Primary Progressive Apraxia of Speech (PPAOS) with graph theory (GT) measures derived from scalp electroencephalography (EEG) recordings. EEGs of 15 agPPA and 7 PPAOS patients were collected during relaxed wakefulness with eyes closed (21 electrodes, 10–20 positions, 256 Hz sampling rate, 1–200 Hz bandpass filter). Eight artifact-free, non-overlapping 1024-point epochs were selected. Via Brainwave software, GT weighted connectivity and minimum spanning tree (MST) measures were calculated for theta and upper and lower alpha frequency bands. Differences in GT and MST measures between agPPA and PPAOS were assessed with Wilcoxon rank-sum tests. Of greatest interest, Spearman correlations were computed between behavioral and network measures in all frequency bands across all patients. There were no statistically significant differences in GT or MST measures between agPPA and PPAOS. There were significant correlations between several network and behavioral variables. The correlations demonstrate a relationship between reduced global efficiency and clinical symptom severity (e.g., parkinsonism, AOS). This preliminary, exploratory study demonstrates potential for EEG GT measures to quantify network changes associated with degenerative speech–language disorders.

BACKGROUND: Parkinson’s disease (PD)-associated psychosis is a well-known non-motor complication, occurring years after diagnosis of PD. Incidence data vary across different studies highlighting a need for long-term observation and clinical definition. OBJECTIVE: To determine the incidence of psychosis in patients with PD and to investigate their survival in an incident cohort study from 1991–2010 in Olmsted County, MN. METHODS: We used the Rochester Epidemiology Project to define an incident-cohort study of parkinsonism (1991–2010) in Olmsted County, MN. A movement-disorder specialist reviewed the electronic medical records and applied diagnosis criteria to PD. Psychosis was diagnosed using of NINDS/NIMH unified criteria. RESULTS: We identified 669 cases of parkinsonism; 297 patients were clinically diagnosed with PD. 114/297 (38.4%) patients had evidence of psychosis (60% male); the median onset age of psychosis was 79.4 years. The incidence of Parkinson’s disease psychosis (PDP) was 4.28/100 person-years. PDP patients had a 71% increased risk of death compared to PD patients. In PD patients without psychosis, men had 73.4% increased risk of death compared to women, whereas no significant sex difference was observed among PDP men vs. women. Of 114 patients diagnosed with psychosis, 59 were treated with antipsychotics. There was no significant difference in survival between treated and untreated patients. CONCLUSION: PDP increased the odds of death compared to PD patients. Men with PD without psychosis had greater odds of death compared to women; however, in PD with psychosis the odds of death were comparable among sexes. Lastly, treatment with anti-psychotics did not significantly affect survival.

AbstractBackgroundDiagnosing Alzheimer’s disease (AD) with biomarkers requires evidence of abnormal beta‐amyloid (Aβ) and tau. Individuals that are Aβ‐positive, but tau‐negative, are considered to have AD neuropathologic change, or early AD. The objective of this study was to determine whether some Aβ‐positive but tau‐negative individuals remain tau‐negative 5‐years later and describe the characteristics of such individuals.MethodWe identified 45 individuals from ADNI who were Aβ‐positive but tau‐negative on PET more than five years after their first abnormal Aβ‐PET (Amyloid+ Low Tau, ALT group). For comparison, we identified 157 Aβ‐positive individuals that had abnormal tau‐PET within five years of their first abnormal Aβ‐PET (i.e. biomarker‐AD). Aβ‐positivity was determined by ADNI‐provided cut‐points of 1.11 for AV45 and 1.08 for florbetapir. A negative tau‐PET scan was defined using Gaussian mixture modeling.ResultThe ALT and biomarker‐AD groups were similar in terms of sex and age at the time of their first Aβ‐PET scan. However, the ALT group was characterized by a lower proportion of APOE ε4 carriers (31% versus 67%, p<0.001), a lower median centiloid (36 versus 85, p<0.001, Figure) and lower likelihood of having a dementia diagnosis at first Aβ PET (2% versus 22%, p<0.001) compared to the biomarker‐AD group. The ALT group also performed better on tests of general cognition and had larger hippocampal volume compared to biomarker‐AD (p<0.001 for all). APOE ε4 status, centiloid and hippocampal volume were all associated with increased odds of elevated tau within 5 years from baseline Aβ‐PET (Table). Within the ALT group, 19 (42%) had high centiloid (>40): these individuals did not differ from the <40 centiloid individuals in clinical diagnosis, but there was some indication of a higher APOE ε4 frequency (42% versus 23%; p=0.30).ConclusionAmyloid‐positive individuals can remain tau‐negative for at least five years, and without becoming demented. Baseline characteristics can help identify these ALT individuals that have a low risk of progression. These findings beg the question of whether more conservative Aβ cut‐points are required for diagnosis of individuals with a high risk of developing AD.

AbstractBackgroundThe apolipoprotein ε4 (APOE4) allele is the most common and well established risk factor for typical late‐onset amnestic Alzheimer’s dementia. Less is known about the role of APOE in patients with atypical clinical presentations of AD, with studies often disagreeing on APOE4 frequencies in these cohorts. We aimed to assess relationships between APOE genotype and age, amyloid‐beta (Aβ) deposition and clinical phenotype in Alzheimer’s disease.Method138 atypical AD patients (57 posterior cortical atrophy and 81 logopenic aphasia) were recruited by the Neurodegenerative Research Group at Mayo Clinic, and underwent APOE testing and Pittsburgh Compound B PET. 318 typical AD patients with APOE genotype and known onset age were identified from ADNI (121 had Aβ‐PET). Aβ‐PET was quantified using the centiloid scale. Logistic regression was used to estimate the proportion of APOE4 carriers by onset age and clinical group. Linear regression was used to estimate mean Aβ centiloid by age, APOE4 status, and clinical group in patients who were Aβ‐positive (centiloid≥19).ResultThe overall frequency of APOE4 carriers was lower in atypical AD (52%) compared to typical AD (66%), particularly for ε4ε4 homozygotes (9% vs 20%). APOE4 frequency increased with age in atypical AD, but showed a bell‐shaped curve in typical AD where highest frequencies were observed between 65‐70 years (Figure). Typical AD showed higher APOE4 frequencies than atypical AD only between the ages of 57 and 69 years. Aβ centiloid was higher in atypical AD compared to typical AD by approximately 16 centiloid units, but did not differ according to APOE4 status (Figure). Aβ centiloid stayed relatively flat with increasing age for both carriers and non‐carriers in atypical AD, but declines with increasing age in typical AD. Relationships between APOE4 and age, as well as centiloid, did not differ between the two atypical AD phenotypes.ConclusionAPOE4 frequency differs between atypical and typical AD but is highly dependent upon age, with higher APOE4 frequency in typical AD only observed between 57 and 69 years. The findings highlight the importance of age and clinical phenotype in understanding the effects of APOE genotype in Alzheimer’s disease.

AbstractBackgroundIndividuals with Primary Progressive Apraxia of Speech (PPAOS) have AOS as the initial and predominant symptom. Many develop aphasia and/or dysarthria later in the disease course. Past research has shown that patients with neurodegenerative AOS experience reduced participation in communication that is further exacerbated by co‐occurring language deficits. Previous studies suggest measures of disease severity do not necessarily correlate with measures of participation restrictions. The aim of this study was to describe changes in communication limitations in these patients over a one‐year interval, measured by 1) the patient via the Communicative Participation Item Bank (CPIB) and 2) the speech‐language pathologist via ASHA's Functional Communication Measures (FCMs) and an adapted Motor Speech Disorder (MSD) severity rating and determine if there are significant changes in these and other objective speech and language measures (e.g. Western Aphasia Battery‐Revised Aphasia Quotient, Apraxia of Speech Rating Scale‐3).MethodSeventeen patients (10 male) with progressive AOS were seen for a second visit approximately one year following the first visit. The measures collected at each visit were statistically compared. Correlations were calculated between the participation ratings and other clinical assessment measures at each visit and qualitatively compared.ResultThere were statistically significant differences in AOS and aphasia severity between visits. There were significant changes in clinical assessments, MSD severity rating, and all ASHA FCMs between visits, but not the CPIB. Comparison of correlations suggest the relationships among clinical and participation measures may change over time.ConclusionThe findings of this study support the use of patient reported outcome measures, like the CPIB, as they may better reflect the patient experience and associated changes in day‐to‐day functioning when other measures may simply index the progression of the disease.

Primary progressive aphasia (PPA) is a progressive language disorder often due to an underlying neurodegenerative disease. The most common pathologies associated with PPA include frontotemporal lobar degeneration (FTLD)‐tau, FTLD‐associated with transactivation response DNA‐binding protein of 43 kDa (TDP‐43) (FTLD‐TDP), and Alzheimer's disease (AD). Accumulating evidence has suggested that Lewy body disease (LBD) can also be associated with PPA. We herein report a 78‐year‐old Caucasian woman who initially presented with levodopa‐responsive parkinsonism at age 67 and later developed cognitive impairment, visual hallucinations, rapid eye movement sleep behavior disorder, and progressive aphasia, characterized by reduced spontaneous speech, word‐finding difficulty, and difficulties in writing and reading. 18Fluorodeoxyglucoase (FDG)‐positron emission tomography (PET) performed at the age of 73 years identified hypometabolism in the frontal (right > left), temporal (left > right), and parietal (left > right) lobes. Neuropathological assessment revealed diffuse LBD (DLBD), AD, and TDP‐43 stage 6 with prehippocampal sclerosis. Senile plaques were numerous, but only a few neurofibrillary tangles were present in the neocortex. The Braak neurofibrillary tangle stage was IV, and the Thal amyloid phase was 3. Lewy‐related pathology was severe in the neocortex, as well as limbic cortices, basal forebrain, amygdala, and brainstem. Compared to 166 DLBD cases with a clinical diagnosis of dementia with Lewy bodies (DLB), the Lewy body count of the patient in this report was highest in the inferior parietal cortex, followed by midfrontal and superior temporal cortices. The findings suggest that severe cortical LBD pathology has contributed to her progressive aphasia. Autopsy cases of LBD presenting as PPA have been reported, but patients with PD and autopsy‐proven DLBD who later developed progressive aphasia have not been reported. Our findings indicate that PD can be associated with progressive aphasia later in the disease course. Although uncommon, LBD should be considered as a differential diagnosis of progressive aphasia.

AbstractBackgroundThe Neurodegenerative Research Group (NRG) at Mayo clinic has followed patients with degenerative speech and/or language disorders for over a decade. In this presentation I will present data from this cohort to inform a broader discussion about the importance of apraxia of speech (AOS) and its subtypes, heterogeneity within the agrammatic PPA variant, and perils of mixed phenotypes.MethodWe reviewed the NRG database to identify patients who initially presented with isolated (PPAOS), AOS in conjunction with agrammatic aphasia (AOS‐PAA), isolated agrammatic aphasia (PAA), and patients who had PAA or AOS overlapping with another degenerative diagnosis. Data from the following clinical instruments were abstracted: WAB AQ (aphasia severity), UPDRS III (parkinsonism), WAB Praxis (ideomotor apraxia), and the MoCA (cognitive ability). We assessed the effect of baseline diagnosis and AOS subtype on these outcomes over time using Bayesian linear mixed effects models. We qualitatively reviewed longitudinal diagnostic trajectories. Preliminary analyses were done on available autopsy data.ResultPAA was associated with greater aphasia at baseline and more rapid decline; PPAOS was associated with normal language at baseline and slow decline; and PAA‐AOS with intermediate baseline aphasia and decline. However, for those with AOS longitudinal decline differed based on AOS subtype: prosodic PPAOS was associated with the slowest decline and phonetic PAA‐AOS with the fastest. Some patients met criteria for non‐PPA disorders during follow up. However, compared to patients who met criteria for these disorders at enrollment, the speech/language‐first patients remained qualitatively distinct, and neither resembled the cases that were mixed at onset. Preliminary autopsy data suggests that AOS is predictive of tau and that the specific tauopathy may be related to the initial syndrome and AOS subtype.ConclusionThe differences between PAA, PAA‐AOS and PPAOS suggest that these disorders should not be lumped. In fact, the longitudinal trajectories are sufficiently different that doing so would add noise to observational and interventional studies. Instead, more detailed phenotyping, including AOS subtype, can aid prognostication and may improve prediction of the underlying pathology. Finally, caution should be exercised when considering PPAOS and PPA patients for other FTLD diagnostic categories.

AbstractBackgroundLogopenic progressive aphasia (LPA) and posterior cortical atrophy (PCA) have distinct patterns of hypometabolism and atrophy. However, these clinical syndromes overlap: visuospatial impairment has been documented in LPA and aphasia in PCA. Rey‐Osterrieth Complex Figure – Copy (Rey‐O Copy) performance is correlated with parietal‐occipital atrophy in LPA. We explore visuospatial impairment on the Rey‐O Copy in LPA and PCA with the Mayo Older Americans Normative Studies (MOANS; which uses Meyers & Meyers criteria) and the Boston Qualitative Scoring System (BQSS). We also examine the proportion of visuospatial impairment in LPA, an under‐appreciated phenomenon, compared to PCA wherein it is fundamental.MethodPatients with LPA (n=25) or PCA (n=23) were recruited by the Neurodegenerative Research Group. Patients underwent comprehensive neuropsychological testing (including Rey‐O Copy) at baseline visit. Cut‐offs of T‐score=40 (BQSS) or scaled score=7 (MOANS) were used based on normative data. Proportions of patients with low performance were examined across LPA and PCA groups. Individual logistic regressions and AUROC analyses were then conducted to quantify differences in visuospatial impairment in LPA and PCA with both scoring systems. Identical analyses were conducted using raw scores corrected for age and sex, to avoid confounds of adjustments in normative data.ResultLPA and PCA groups were comparable on age, sex, and disease duration (LPA Median = 2.7 years, IQR=1.9‐3.5; PCA Median = 3.6 years, IQR=1.5‐4.9); p=.823). Surprisingly, 52‐72% of LPA participants demonstrated impaired performance across scoring systems, relative to 96% of the PCA group identified by both scoring systems. Odds ratios (OR) for the Rey‐O Copy MOANS (OR = 1.59; p=.008) and BQSS Copy Presence and Accuracy index (CPA; OR = 1.13, p=.002) were significant, persisting for age‐ and sex‐ controlled raw scores, suggesting the odds of a PCA diagnosis increased with decrement on Rey‐O Copy. AUROC values ranged from 0.77 to 0.88 across these indices.ConclusionA substantial number of LPA participants showed visuospatial impairment, despite initial primary language complaints. Differentiation of diagnostic groups by the Rey‐O is likely underscored by relatively uniform, profound, visuospatial impairment in PCA with more variability in LPA. However, the degree of visuospatial impairment overlaps remarkably across diagnoses.

AbstractBackgroundOptimizing software to measure Flortaucipir (AV‐1451) tau PET SUVR change‐over‐time is crucial for longitudinal observational studies and potential trials of anti‐tau agents. Ideal measurements should have good repeatability (low noise), and should clearly separate clinical cognitive groups.MethodWe compared 456 software methods varying in their use of: two major packages, each with cross‐sectional and longitudinal variants (FreeSurfer and an in‐house pipeline using SPM12, MCALT, and ANTs); three temporal‐lobe target regions (each with/without adjacent WM); seven reference regions; five PVC methods; and two statistics (mean vs. median). We used linear mixed‐effects models to compare methods using scans from 88 Mayo Clinic participants (46 cognitively impaired, 42 impaired, excluding those with symptoms inconsistent with the AD pathology imaged by Flortaucipir), each with three visits with PET and MRI. From these models, we calculated two criteria for each pipeline: relative measurement error (residual from the linear mixed model), and the t‐statistic for testing differences in group‐wise rates of change (which has a Cohen’s d interpretation of effect size for longitudinal data). We plotted both criteria orthogonally, then ranked methods overall according to the average of their ranks on both individual criteria.ResultMethods using the entorhinal cortex target region performed poorly in both criteria (Figure 1). The top 10 methods by the combined criteria (top‐left in Figure 2) all used the Mayo SPM12‐based pipeline with an eroded supratentorial WM (eSWM) or composite (eSWM + pons + cerebellum) reference region. Among them, 9/10 used the longitudinal variant, 8/10 used 2‐compartment PVC, and 7/10 used median. Altogether, our results favored SUVR methods with a longitudinal design (simultaneously calculating all time‐points), a temporal‐lobe meta‐ROI including adjacent WM, a composite reference region, 2‐class voxel‐based (Meltzer) PVC, and median statistics. Methods using GTM‐Tau PVC (Baker et al., Data in Brief, 2017) had the highest group separation but relatively large measurement error, giving them ranks 197 and 199 out of 456 with the joint (average rank) criteria.ConclusionCompared with popular FreeSurfer‐based methods (ranks 176, 258; Figure 2), our top‐ranked approaches approximately halved error in SUVR measurements while maintaining or improving group separation.