Kremers, Walter K.

Introduction: An artificial intelligence (AI) algorithm applied to electrocardiography (ECG) during normal sinus rhythm (NSR) has been shown to predict concomitant atrial fibrillation (AF). We sought to characterize the value of AI-ECG as a predictor of future AF and assess its performance compared to other clinical prediction scores in a population-based sample. Methods: We calculated the AI-ECG probability during NSR in patients who enrolled in the population-based Mayo Clinic Study of Aging with at least one ECG in NSR within two years prior and no history of AF at the time of the baseline study visit. The cumulative incidence of AF was estimated for strata defined by AI-ECG probability and CHARGE-AF score. Cox proportional hazards were fit to assess the independent prognostic value and interaction of AI-ECG probability and CHARGE-AF score. Concordance (c) statistics were calculated for AI-ECG probability, CHARGE-AF score and combined AI-ECG and CHARGE-AF score. Results: A total of 1,936 patients with a median age 75.8 (quartile range [QR] 70.4, 81.8) years, median CHARGE-AF score 14.0 (QR 13.2, 14.7) and median CHADS2VASC score 3 (QR 2, 4) were included in the analysis. The cumulative incidence of AF increased in a stepwise fashion across quartiles of AI-ECG probability and CHARGE-AF score (Figure 1). When compared in the same model, both AI-ECG probability (hazard ratio [HR] 1.76, 95% confidence interval [CI] 1.51-2.04) and CHARGE-AF score (HR 1.90, 95% CI 1.58-2.28) independently predicted AF without significant interaction (p=0.54). C statistics were 0.69 (95% CI 0.66-0.72) for AI-ECG probability, 0.69 (95% CI 0.66-0.71) for CHARGE-AF and 0.72 (95% CI 0.69-0.75) for combined AI-ECG and CHARGE-AF score. Conclusions: In the Mayo Clinic Study of Aging, both the AI-ECG probability and CHARGE-AF score independently predicted time to AF. The AI-ECG may offer a means to assess risk with a single test without requiring manual or automated clinical data abstraction.

Background and Purpose: Cerebral microbleeds (CMBs) are represented by small areas of hemosiderin deposition, detected on brain magnetic resonance imaging (MRI), and found in ≈23% of the cognitively normal population over age of 60 years. CMBs predict risk of hemorrhagic and ischemic stroke. They correlate with increased cardiovascular mortality. In this article, we sought to determine in a population-based study whether antithrombotic medications correlate with CMBs and, if present, whether the association was direct or mediated by another variable. Methods: The study consisted of 1253 participants from the population-based Mayo Clinic Study of Aging who underwent T2* gradient-recalled echo magnetic resonance imaging. We tested the relationship between antithrombotic medications and CMB presence and location, using multivariable logistic-regression models. Ordinal logistic models tested the relationship between antithrombotics and CMB frequency. Using structural equation models, we assessed the effect of antithrombotic medications on presence/absence of CMBs and count of CMBs in the CMB-positive group, after considering the effects of age, sex, vascular risk factors, amyloid load by positron emission tomography, and apoE. Results: Two hundred ninety-five participants (26.3%) had CMBs. Among 678 participants taking only antiplatelet medications, 185 (27.3%) had CMBs. Among 95 participants taking only an anticoagulant, 43 (45.3%) had CMBs. Among 44 participants taking an anticoagulant and antiplatelet therapy, 21 (48.8%) had CMBs. Anticoagulants correlated with the presence and frequency of CMBs, whereas antiplatelet agents were not. Structural equation models showed that predictors for presence/absence of CMBs included older age at magnetic resonance imaging, male sex, and anticoagulant use. Predictors of CMB count in the CMB-positive group were male sex and amyloid load. Conclusions: Anticoagulant use correlated with presence of CMBs in the general population. Amyloid positron emission tomography correlated with the count of CMBs in the CMB-positive group.

BackgroundTo operationalize the National Institute on Aging – Alzheimer's Association (NIA‐AA) Research Framework for Alzheimer's Disease 6‐stage continuum of clinical progression for persons with abnormal amyloid.MethodsThe Mayo Clinic Study of Aging is a population‐based longitudinal study of aging and cognitive impairment in Olmsted County, Minnesota. We evaluated persons without dementia having 3 consecutive clinical visits. Measures for cross‐sectional categories included objective cognitive impairment (OBJ) and function (FXN). Measures for change included subjective cognitive impairment (SCD), objective cognitive change (ΔOBJ), and new onset of neurobehavioral symptoms (ΔNBS). We calculated frequencies of the stages using different cutoff points and assessed stability of the stages over 15 months.ResultsAmong 243 abnormal amyloid participants, the frequencies of the stages varied with age: 66 to 90% were classified as stage 1 at age 50 but at age 80, 24 to 36% were stage 1, 32 to 47% were stage 2, 18 to 27% were stage 3, 1 to 3% were stage 4 to 6, and 3 to 9% were indeterminate. Most stage 2 participants were classified as stage 2 because of abnormal ΔOBJ only (44–59%), whereas 11 to 21% had SCD only, and 9 to 13% had ΔNBS only. Short‐term stability varied by stage and OBJ cutoff points but the most notable changes were seen in stage 2 with 38 to 63% remaining stable, 4 to 13% worsening, and 24 to 41% improving (moving to stage 1).InterpretationThe frequency of the stages varied by age and the precise membership fluctuated by the parameters used to define the stages. The staging framework may require revisions before it can be adopted for clinical trials. ANN NEUROL 2021;89:1145–1156

AbstractBackgroundThe medium‐ to long‐term outcomes of living kidney donors with hypertension compared to normotensive donors are not well understood, especially with the recent changes in hypertension guidelines.MethodsWe studied a cohort of 950 living kidney donors using different definitions of hypertension based on either ≥140/90 or ≥130/80 mmHg thresholds and based on either office or ambulatory blood pressure readings. Microstructural features on kidney biopsy at the time of donation were compared using different definitions of hypertension.ResultsAfter adjusting for years of follow‐up, age, sex, and baseline eGFR, hypertension (by any definition) did not significantly predict an eGFR < 45 ml/min/1.73 m2 at a median follow‐up of 10 years postdonation, though there was a borderline association with ambulatory blood pressure ≥ 130/80 mmHg predicting a 40% decline in eGFR (OR = 1.53, 1.00–2.36; p = .051). Proteinuria was predicted by office blood pressure ≥ 140/90 mmHg and by nondipper profile on nocturnal ambulatory blood pressure measurements. At the time of donation, larger glomeruli and arterial hyalinosis on biopsy were associated with hypertension defined by either ≥140/90 or ≥130/80 mmHg (by office or ambulatory measurements). Nocturnal nondipper status was associated with larger glomeruli size but not arteriolar hyalinosis when compared to dippers.ConclusionsIn programs that accept donors with controlled hypertension, various definitions of hypertension are associated with histological findings in the donated kidney, but none predict a clinically significant decline in kidney function 10 years after donation. These data support allowing healthy individuals with controlled hypertension to donate a kidney. However, donors with office hypertension (≥140/90 mmHg) and nondippers (regardless of hypertension status) are at greater long‐term risk for proteinuria, and particularly for these donors, longer follow‐up is warranted.

Abstract Objective Neuropsychiatric symptoms (NPS) are associated with the risk of incident mild cognitive impairment (MCI) and dementia. We examined associations between NPS and the outcomes of global and domain‐specific cognitive trajectories. Methods In this longitudinal study conducted in the setting of the population‐based Mayo Clinic Study of Aging, 5081 community‐dwelling, nondemented individuals aged ≥50 years (51% males) underwent NPS assessment using Neuropsychiatric Inventory Questionnaire (NPI‐Q), and Beck Depression and Anxiety Inventories (BDI‐II, BAI). Global and domain‐specific (memory, language, attention, and visuospatial skills) cognitive performance was assessed through neuropsychological testing every 15 months. Associations between baseline NPS and trajectories for individual yearly change in cognitive z‐scores were calculated using linear mixed‐effect models. Results Cognition declined regardless of NPS status over the median follow‐up of 4.5 years. Presence of NPS was associated with increased cognitive decline. Differences in annualized change in global cognition z‐scores for participants with NPS compared to without NPS ranged from −0.018 (95% CI −0.032, −0.004; p = 0.011) for irritability to −0.159 (−0.254, −0.065; p = 0.001) for hallucinations. Associations between NPS and annual decline in global cognition were significant for most NPI‐Q‐assessed NPS and clinical depression (BDI‐II≥13). Participants with NPI‐Q‐assessed depression, apathy, nighttime behavior, and clinical depression had greater decline in all domain‐specific z‐scores; presence of delusions and anxiety was associated with more pronounced decline in language, attention and visuospatial skills. Conclusion NPS were associated with a more accelerated cognitive decline. Clinical assessment and potential treatment of NPS is warranted even in a community setting as NPS may impact cognitive decline in nondemented individuals.

Background: Atrial fibrillation (AF) is an established risk factor for ischemic stroke, but it can be paroxysmal and may go undiagnosed. An artificial intelligence (AI)-enabled ECG acquired during normal sinus rhythm was recently shown to detect silent AF. The objective of this study was to determine if AI-ECG AF score is associated with presence of cerebral infarcts. Methods: Participants from a population-based study ages 30 to 95 years with T2 fluid attenuation inversion recovery (FLAIR) MRI obtained between October 10, 2011, and November 2, 2017 were considered for inclusion. Participants without ECG were excluded. AI-ECG score was calculated using most recent ECG with normal sinus rhythm at the time of MRI. Presence of infarcts was determined on FLAIR MRI scans. Logistic regression was run to evaluate the relationship between AI-ECG AF score and presence of cerebral infarcts. Similar analyses were performed using history of AF rather than AI-ECG AF score as predictor. Age and sex were included as covariates. We also examined whether a high-threshold AI-ECG score was associated with infarcts. In a prior study, an AI-ECG AF score > 0.5 was associated with a cumulative incidence of AF of 21.5% at 2 years and 52.2% at 10 years. Results: This study included 1,373 individuals. Average age was 69.6 years and 53% of participants were male. There were 136 (10%) individuals with ECG-confirmed AF; 1237 (90%) participants had no AF history. Of participants with AF, 23% (n=31) were on anticoagulation, 47% (n=64) were on antiplatelet and 18% (n=24) were on dual therapy. Only 1.3% (n=16) of patients without AF were on anticoagulation and 47% (n=578) were on antiplatelet therapy. Ischemic infarcts were detected in 214 (15.6%) patients. As a continuous measure AI-ECG was associated with infarcts but not after adjusting for age and sex (p=0.46). AI-ECG AF score > 0.5 was associated with infarcts ( p < 0.001); even after adjusting for age and sex ( p = 0.03). History of AF was also associated with infarcts after adjusting for age and sex ( p = 0.018). Conclusion: AI-ECG AF score and history of AF were associated with presence of cerebral infarcts after adjusting for age and sex. This tool could be useful in select patients with cryptogenic stroke but further investigation would be required.

AbstractIntroductionThis study evaluated the diagnostic accuracy of the Cogstate Brief Battery (CBB) for mild cognitive impairment (MCI) and prodromal Alzheimer's disease (AD) in a population‐based sample.MethodsParticipants included adults ages 50+ classified as cognitively unimpaired (CU, n = 2866) or MCI (n = 226), and a subset with amyloid (A) and tau (T) positron emission tomography who were AD biomarker negative (A–T–) or had prodromal AD (A+T+).ResultsDiagnostic accuracy of the Learning/Working Memory Composite (Lrn/WM) for discriminating all CU and MCI was moderate (area under the curve [AUC] = 0.75), but improved when discriminating CU A–T– and MCI A+T+ (AUC = 0.93) and when differentiating MCI participants without AD biomarkers from those with prodromal AD (AUC = 0.86). Conventional cut‐offs yielded lower than expected sensitivity for both MCI (38%) and prodromal AD (73%).DiscussionClinical utility of the CBB for detecting MCI in a population‐based sample is lower than expected. Caution is needed when using currently available CBB normative data for clinical interpretation.

AbstractBackgroundFamilial frontotemporal dementia (f‐FTLD) is typically caused by mutations in one of three genes: microtubule‐associated protein tau (MAPT), progranulin (GRN), and a repeat expansion in the chromosome 9 open reading frame 72 (C9orf72) gene. Accurate characterization of the natural history of each mutation is important for clinical prognostication and clinical trial design, and it could shed light on disease biology. Such models have not been thoroughly developed using participants that represent all disease stages, with longitudinal data. We characterized the trajectory of atrophy in each gene by using longitudinal voxel‐wise analyses of gray matter volume, and we assessed whether functional independence declined in tandem.MethodF‐FTLD participants (n=100) with a known mutation (MAPT+ (n=28), GRN+ (n=33), C9orf72+ (n=39)) were grouped according to disease stage (CDR®+NACC FTLD module). We included participants with at least two structural MRIs at a given disease stage: presymptomatic (CDR®+NACC‐FTLD=0, n=57), mild/questionable (CDR®+NACC‐FTLD=0.5, n=15), and symptomatic (CDR®+NACC‐FTLD ≥1, n=28). We fitted longitudinal linear mixed effects models to extract mean atrophy rates in each lobe compared to longitudinal imaging from family members without mutations (n=60). All results presented below were significant at p<.001.ResultUsing the left frontal lobe as an exemplar, in the presymptomatic stage, MAPT mutation carriers showed the greatest rate of atrophy compared to controls (88 mm3/year more volume loss than controls), followed by GRN+ (65 mm3/year) and then C9orf72+ (49 mm3/year). In the mild/questionable stage, MAPT+ showed a greater divergence from controls (374 mm3/year) than did GRN+ (107 mm3/year) or C9orf72+ (223 mm3/year). In the symptomatic stage, MAPT+ again lost volume at the fastest rate (2,099 mm3/year), followed by GRN+ (1,360 mm3/year). C9orf72 expansion carriers showed a much slower rate of volume loss (115 mm3/year). Similar patterns were observed for other brain regions. In contrast to the imaging results, C9orf72+ exhibited similar rates of functional decline compared to GRN+ and MAPT+ at all levels of disease severity.ConclusionThe primary f‐FTLD genes show divergent atrophy trajectories as a function of disease stage, with C9orf72 expansion carriers exhibiting a slow degeneration throughout the disease course, possibly due to unique pathophysiological mechanisms.

AbstractBackgroundData shared for research use must be de‐identified to protect participant confidentiality. A recent study has shown that face imagery from brain imaging studies could be used to re‐identify anonymous participants via automatic face recognition, linking them with all study‐collected protected health information. Here we measure how well existing software for removing MRI face imagery protects against this privacy threat.MethodWe recruited 153 participants with existing 3D FLAIR MRI and photographed their face from 5 varying angles. We used surf_ice to reconstruct each participant’s face from their MRI and tested whether free, publicly‐available face recognition software (Microsoft Azure) could match their photos to their correct MRI. We tested two existing programs for removing face voxels in MRI: mri_deface (FreeSurfer) and pydeface (python package), and one in‐house implementation. We observed that existing methods often left participants’ eyes intact while removing most of their nose/mouth/chin. This prevented Azure from detecting the reconstructions as faces, therefore preventing its attempt of face recognition. We hypothesized that the retained eye imagery alone could be sufficient for face recognition, if a face were detected. Therefore, we also reconstructed each de‐faced MRI after applying novel pre‐processing that replaced any removed/missing face voxels with those from our population‐average template, e.g. replacing missing noses/mouths/chins with population‐average noses/mouths/chins, before testing face recognition.ResultsWithout any MRI de‐facing, Microsoft Azure correctly matched 78% of participants’ photos to their MRI. After mri_deface and pydeface, 8% and 6% were still identified, respectively. After preprocessing the de‐faced MRI to replace missing areas with those from a population‐average template prior to face reconstruction, the face imagery remaining in the de‐faced images was enough to allow re‐identification of 23% and 20% of participants, respectively. With an in‐house face removal implementation that removes the whole face from every image, 0% of faces were re‐identified from the standard reconstructions, and 5% were identified from the “missing‐parts‐replacement” reconstructions.ConclusionsEven after applying popular MRI de‐facing software, Microsoft Azure face recognition automatically re‐identified up to 23% of participants. Better de‐facing software is needed to remove more of the face more reliably, without affecting brain measurements from de‐faced images.