Citations

AbstractBackgroundSubjective cognitive concerns are common in functionally intact adults, potentially indicating future cognitive decline. Remote smartphone cognitive testing holds promise for objectively tracking cognition in individuals reporting complaints. In our initial exploration of the link between subjective cognitive complaints and digital clinical outcomes, we examined participants' self‐reported cognitive complaints' association with smartphone tests on memory and executive functioning.MethodsParticipants, without self‐reported MCI or dementia, were recruited from the UCSF Brain Healthy Registry (BHR) to complete cognitive tasks on their personal smartphone with the ALLFTD Mobile App. Participants also completed the 12‐item Everyday Cognition Screening Tool (ECog‐12), assessing subjective cognitive impairment, online as part of BHR. Average ECog‐12 scores, including Memory and Executive Functioning subdomains, were calculated. Correlations between scores from online ECog‐12 scores and app‐based smartphone tasks were examined. Regression analyses, adjusting for age, education, and sex, were performed to evaluate the relationship between ECog‐12 and smartphone tasks (Stroop & Flanker) measuring executive function and memory.ResultsOut of 2,283 invited BHR participants, 456 (mean age: 64.6 years (SD = 11.8) and education: 16.3 years (SD = 2.4)) completed smartphone tasks (81.1% completion rate). Age was negatively associated with executive functioning and memory tasks (ß’s = ‐0.51 to ‐0.54, p’s<0.05), while higher education correlated with better performance (ß’s = 0.10‐0.14, p’s<0.05). Sex was not significantly associated with performance on any task (ß’s = ‐0.04 to 0.09, p’s>0.05). Subjective cognitive concerns were linked to tasks of executive functioning (Stroop: ß = ‐0.18, p = 0.002) and memory (ß = ‐0.14, p = 0.02). Endorsement of executive functioning items on ECog‐12 associated with Stroop performance (ß = ‐0.16, p = 0.005), and memory‐specific complaints were related to memory performance (ß = ‐0.1, p = 0.09). The relationship with flanker scores was not significant (ß = ‐0.08, p = 0.17). Similar patterns emerged with the online ECog‐12; online and app‐based scores are strongly correlated (ß = 0.75, p<0.05).ConclusionsIn this cognitively intact sample, subjective cognitive concerns were linked to poorer performance on ALLFTD Mobile App tasks assessing executive functioning and memory. These findings suggest that unsupervised remote smartphone testing merits further consideration to monitor cognition in individuals with subjective cognitive concerns. Subsequent studies in deeply phenotyped cohorts with follow‐up assessments are essential for understanding the clinical relevance of these results.

AbstractBackgroundFew normative data for computerized measures administered in unsupervised remote environments are available. We aimed to determine what variables to include in normative models for remote self‐administered assessments, develop normative data for measures administered through Mayo Test Drive (MTD, a multi‐device remote cognitive assessment platform) and evaluate application of norms.Method1240 adults ages 33‐100 (96% White) from the Mayo Clinic Study of Aging and Mayo Alzheimer’s Disease Research Center met normative sample inclusion criteria that included a concordant Cognitively Unimpaired (CU) diagnosis (3 independent raters all diagnosed CU) and CDR = 0 (see Table 1 for sample characteristics). Raw test scores were converted to normalized scaled scores. We derived regression‐based normative data that adjusted for age, age2, sex and education (base model); alternative models are also provided (age+age2+sex; age+ age2). We evaluated whether additional model terms beyond the base model were needed using an a priori cut‐off of 1% variance improvement in adjusted R2 over and above the base model. We examined rates of low test performance defined as performance below ‐1 SD in independent validation samples (n = 167 concordant CU, n = 64 mild cognitive impairment (MCI), n = 14 dementia). Rates are considered significantly different than expected when 95% confidence intervals (CIs) do not include the expected 14.7% base rate frequency.ResultNo additional potential model terms met our a priori cut‐off for inclusion beyond the base model (see Table 2), including other quadratic and cubic terms or interactions, device type, response input type, potential session interference, and national area deprivation index (ADI available in subset; N = 776). Application of norms showed expected rates of low test performance among CU participants (95% CI includes expected base rate) and significantly greater than expected rates of low test performance for all variables in MCI (e.g., 62.5% for composite) and dementia (e.g., 71.4% for composite; see Table 3).ConclusionTypical normative models appear appropriate for remote self‐administered measures and are sensitive to cognitive impairment. Device type and response input did not explain enough variance to warrant inclusion in normative models but are important for individual level interpretation. Future work will improve representation of individuals from under‐represented groups.

Abstract Attempts to activate an anti-tumor immune response in glioblastoma (GBM) have been met with many challenges due to its inherently immunosuppressive tumor microenvironment. The degree and mechanisms by which molecularly and phenotypically diverse tumor-propagating glioma stem cells (GSCs) contribute to this state are poorly defined. In this study, our multifaceted approach combining bioinformatics analyses of clinical and experimental datasets, single-cell sequencing, and molecular and pharmacologic manipulation of patient-derived cells identified GSCs expressing immunosuppressive effectors mimicking regulatory T cells (Tregs). We show that this I mmunosuppressive T reg- L ike (ITL) GSC state is specific to the mesenchymal GSC subset and is associated with and driven specifically by TGF-β type II receptor (TGFBR2) in contrast to TGFBR1. Transgenic TGFBR2 expression in patient-derived GBM neurospheres promoted a mesenchymal transition and induced a 6-gene ITL signature consisting of CD274 (PD-L1), NT5E (CD73), ENTPD1 (CD39), LGALS1 (galectin-1), PDCD1LG2 (PD-L2), and TGFB1. This TGFBR2-driven ITL signature was identified in clinical GBM specimens, patient-derived GSCs and systemic mesenchymal malignancies. TGFBR2 High GSCs inhibited CD4+ and CD8+ T cell viability and their capacity to kill GBM cells, effects reversed by pharmacologic and shRNA-based TGFBR2 inhibition. Collectively, our data identify an immunosuppressive GSC state that is TGFBR2-dependent and susceptible to TGFBR2-targeted therapeutics.

AbstractBackgroundWe recently developed a novel tau‐PET summary measure THETA, capturing regional heterogeneity and identifying tau status, using ground truth visual assessments from a large single‐center cross‐sectional dataset and validated on independent cohorts [1, 2]. In this study, we aimed to evaluate the performance of THETA on longitudinal and histopathology data.MethodWe included longitudinal tau‐PET ([18F]flortaucipir) data from 696 Mayo Clinic Study of Aging (MCSA) and ADRC participants, with histopathology in n = 90. Fig. 1 shows the model that uses regional standard uptake value ratios (SUVR) and a target of binary class of tau positivity for prediction of THETA. This model was applied to predict tau status on each followup scan. Slopes of the meta‐ROIs’ tau SUVRs and THETA were evaluated by diagnostic group and as a function of baseline amyloid SUVR in discordant CU participants (where meta‐ROI and visual assessments did not match at baseline) and in incident‐THETA+ CU participants (whose THETA moved from below to above 1 over serial tau‐PET scans). We evaluated tau measurements as a function of Braak stages for neurofibrillary tangles.ResultLongitudinal plots of meta‐ROIs for diagnostic groups were very similar, but the expanded range of THETA based on visual positivity prediction clearly identified tau positive or tau‐negative scans. In discordant‐CU (n = 97) and incident‐THETA+ CU participants (n = 14 all amyloid positive at follow‐up but only 65% at baseline), the relationship between baseline amyloid and rate of tau increase was stronger for THETA than temporal meta‐ROI (Fig. 2). Separation between baseline and follow‐up was greater for THETA (t‐statistics = 90) compared to temporal meta‐ROI (t‐statistics = 20) (p<0.01) in the incident‐THETA+ CU participants. THETA showed a slightly stronger association with Braak stage than meta‐ROIs (rho = 0.87 vs. ≤0.83, p<0.05), with better separation of clinical diagnoses (Fig. 3).ConclusionTHETA remained clearly negative or positive in MCI and AD, providing consistent information on underlying etiology of impairment at both baseline and follow‐up. Although binary in its construction, THETA both provided separation of values based on tau status and its change correlated with baseline amyloid burden especially in discordant‐CU where tau deposition is not in typical meta‐ROIs. Further work is needed to confirm if THETA captures early tau changes.

AbstractBackgroundBiomarker cutpoints have received significant importance in AD research. Reliable worsening method based on serial biomarker data has been proposed (Jack et al., 2017 Alz & Dem). Advanced machine learning models (ML) and explainable AI provide an opportunity to enhance existing techniques. We aimed to estimate biomarker cutpoints on cross‐sectional data using ML by identifying important transitions in the biomarker values.MethodWe first trained an ML model with the biomarker of interest as an input including other available biomarkers, demographics, and diagnosis, and predicted global cognition z‐score. We then obtained SHAP values (Shapley Additive exPlanations), fit a curve between the SHAP values and biomarker values, and calculated the first derivatives (slope) of the curve. Then, a CUSUM (Cumulative Sum) analysis was used to identify SHAP change points and cross‐referenced to values where the slopes were steepest near the zero‐crossing location of the curve. A zero SHAP is an important indicator of a feature’s progressive importance for model prediction because a feature’s influence on prediction significantly shifts when its SHAP values go from positive to negative or vice versa. However, relying on zero‐crossing alone may be insufficient when the relationship is non‐linear. Consequently, we calculated the second derivatives to identify inflection points (peaks and troughs) along the curve. Lastly, we compiled all the potential cutpoints and performed post‐processing as shown in Fig. 1.ResultThe algorithm identified cutpoints at 1.44 SUVR for all and at 1.42 SUVR for CU participants (Fig. 2). At age 70, both cutpoints estimated the percentage of abnormals at less than 40% (Fig. 3).ConclusionTPE amyloid‐PET SUVR cutpoints for CUs equal the reported RW values (1.42 SUVR). A strength of the TPE is its application on cross‐sectional data making it versatile for various biomarkers across different cohorts.

AbstractBackgroundFrontotemporal dementia (FTD) is a diverse spectrum of neurodegenerative disorders impacting the frontal and temporal lobes, primarily manifesting as behavioral and/or language issues. Accurate diagnosis is crucial for effective management, potential treatment, and care planning. While speech analysis has shown promise in detecting cognitive markers of Alzheimer’s disease (AD), its exploration in FTD is limited. This study examined if linguistic features from remote, app‐based speech collection can differentiate healthy controls (HC) from individuals with behavioral variant FTD (bvFTD), the most common FTD variant.MethodParticipants included 15 bvFTD (mean age = 64.3, SD = 9.1 years) and 19 HC (mean = 55.8, SD = 14.2) from the ALLFTD Consortium. Participants were prompted to talk freely about a happy memory; audio recordings were collected with the ALLFTD Mobile App, transcribed using OpenAI’s large Whisper model, and manually quality‐checked. We extracted 98 linguistic features using our in‐house pipeline (Fig 1). Features were selected for classification analysis if they differed between groups based on a MANCOVA adjusted for age, sex/gender, and education, with feature reduction in case of collinearity. We then determined if these features could classify groups using multi‐variable logistic regression models, and subsequently calculated its ROC‐AUC score to measure sensitivity/specificity.ResultBased on the MANCOVA, three features were selected for classification analysis: average seconds of non‐speech (i.e., silence), ratio of number of complex nominal clauses to total clauses, and number of restarts (i.e., disfluency). The logistic model showed that this set could distinguish between groups (Omnibus p = .001; Nagelkerke R2 = .499), and their saved prediction probabilities indicated a ROC‐AUC score of 84.8% accuracy (Fig 2).ConclusionLeveraging a flexible speech capture app with remote accessibility and state‐of‐the‐art automated speech analysis, we identified linguistic features that can distinguish HC from bvFTD. This approach presents a promising screening tool for detecting bvFTD, offering quick, easily administered assessments with the potential for scalable implementation outside of clinical settings. Next steps are to investigate linguistic changes in presymptomatic mutation carriers and compare speech profiles between FTD and AD. These results hold meaningful implications for addressing current challenges in FTD therapeutics, potentially offering sensitive outcome measures for clinical trials and future tools for personalized cognitive assessment and long‐term monitoring.

AbstractBackgroundFrontotemporal dementia (FTD) is a common young‐onset dementia. Challenges to in‐person FTD evaluations (e.g., behavioral symptoms, disease rarity), highlight the need to develop remote, low‐burden assessment techniques. A growing literature supports passive digital phenotyping for monitoring neurobehavioral change. Thus, we examined the utility of passively collected data from smartphones to detect prodromal or symptomatic FTD compared to that of the Montreal Cognitive Assessment (MoCA), a common cognitive screener.Method199 adults enrolled in the ALLFTD Mobile App study (mean age = 53.4 [SD = 15.2]; 58% women) completed up to 6 months of passive smartphone monitoring via the ALLFTD Mobile App. 55% were unimpaired [CDR®+NACC FTLD = 0], 22% had prodromal FTD [CDR®+NACC FTLD = 0.5], and 23% had symptomatic FTD [CDR®+NACC FTLD≥1] with a variety of syndromic presentations. Battery percentage was collected frequently (median interval = 45 min), and total daily battery usage was calculated as a proxy for smartphone use (higher battery usage = more smartphone use). Daily step counts were also passively collected. Logistic regression classified prodromal or symptomatic FTD vs. unimpaired as a function of: MoCA score [Model 1]; 7 passive features characterizing average, inter‐day variability, and changes over time in smartphone use and movement [Model 2]; or passive smartphone features and MoCA [Model 3]. Analyses were repeated excluding participants with symptomatic FTD to detect prodromal FTD vs. unimpaired.ResultThe area under the curves (AUCs) for detecting prodromal or symptomatic FTD (vs. unimpaired) from the MoCA alone [Model 1] and passive smartphone features alone [Model 2] were 0.77 (95%CI: 0.70‐0.85) and 0.77 (95%CI: 0.70‐0.84), respectively. Their combination [Model 3] resulted in a significantly improved AUC of 0.86 (95%CI: 0.80‐0.92) compared to Model 1 (p = 0.001). Significant Model 3 predictors included average daily battery usage (p = 0.02), slopes of change in step count (p = 0.04), and MoCA score (p<0.01). When distinguishing prodromal FTD from unimpaired, passive smartphone features alone (AUC = 0.75) were significantly better than the MoCA alone (AUC = 0.66; p<0.01), and their combination had the largest AUC (0.80).ConclusionResults support the utility of novel passively collected information from smartphones for detecting and monitoring FTD without contributing to assessment burden. With continued validation, passive digital monitoring methodologies have potential to increase access to dementia care.

AbstractBackgroundPrior longitudinal studies among older adults have documented associations between hearing loss and changes in brain morphology. Whether interventions involving hearing aids can reduce age‐related atrophy is unknown. A substudy within the Aging and Cognitive Health Evaluation in Elders (ACHIEVE, Clinicaltrials.gov Identifier: NCT03243422) randomized controlled trial tested the effect of a best‐practices hearing intervention versus health education control on three‐year change in cortical thickness among older adults with hearing loss.MethodThe ACHIEVE study enrolled 977 community‐dwelling adults aged 70‐84 years at baseline (2018‐2019) with untreated hearing loss (better ear pure tone average [0.5‐4 kHz] ≥30 and <70 dB HL) and without substantial cognitive impairment from four sites across the U.S. (Jackson, MS, Forsyth County, NC, Minneapolis, MN, Washington County, MD). Participants were randomized to a hearing intervention (provision of hearing aids and related technologies, counseling, and education) or a health education control (individual sessions with a health educator covering topics relevant to chronic disease and disability prevention). Three‐dimensional magnetic resonance imaging was performed on 3 Tesla Siemens scanners in a subsample of 445 participants at the ACHIEVE baseline and three‐year follow‐up. Linear mixed effects models were used in intention‐to‐treat analyses to estimate three‐year change in cortical thickness. All models adjusted for baseline measures of hearing loss, recruitment source, site, age, sex, and education. Missing outcome and covariate data was imputed to mitigate bias caused by informative attrition.ResultAt baseline, 224 participants were women (50.3%), 52 participants were Black (11.7%), and the mean (SD) age was 76.4 (4.0) years old (Table 1). Compared to the health education control, the hearing intervention exhibited a nominally protective effect on three‐year change in average cortical thickness (Figure 1). The greatest effect size for cortical thickness was observed in the occipital lobe, while the smallest effect size was detected in the temporal lobe. Statistically significant effects were detected in the pars orbitalis, rostral anterior cingulate, posterior cingulate, and isthmus cingulate (Figure 2).ConclusionHearing aid use may reduce decline in cortical thickness among older adults. The effects of hearing aids may be greatest in regions other than those associated with the auditory cortex.

AbstractBackgroundFew normative data for computerized measures administered in unsupervised remote environments are available. We aimed to determine what variables to include in normative models for remote self‐administered assessments, develop normative data for measures administered through Mayo Test Drive (MTD, a multi‐device remote cognitive assessment platform) and evaluate application of norms.Method1240 adults ages 33‐100 (96% White) from the Mayo Clinic Study of Aging and Mayo Alzheimer’s Disease Research Center met normative sample inclusion criteria that included a concordant Cognitively Unimpaired (CU) diagnosis (3 independent raters all diagnosed CU) and CDR = 0 (see Table 1 for sample characteristics). Raw test scores were converted to normalized scaled scores. We derived regression‐based normative data that adjusted for age, age2, sex and education (base model); alternative models are also provided (age+age2+sex; age+ age2). We evaluated whether additional model terms beyond the base model were needed using an a priori cut‐off of 1% variance improvement in adjusted R2 over and above the base model. We examined rates of low test performance defined as performance below ‐1 SD in independent validation samples (n = 167 concordant CU, n = 64 mild cognitive impairment (MCI), n = 14 dementia). Rates are considered significantly different than expected when 95% confidence intervals (CIs) do not include the expected 14.7% base rate frequency.ResultNo additional potential model terms met our a priori cut‐off for inclusion beyond the base model (see Table 2), including other quadratic and cubic terms or interactions, device type, response input type, potential session interference, and national area deprivation index (ADI available in subset; N = 776). Application of norms showed expected rates of low test performance among CU participants (95% CI includes expected base rate) and significantly greater than expected rates of low test performance for all variables in MCI (e.g., 62.5% for composite) and dementia (e.g., 71.4% for composite; see Table 3).ConclusionTypical normative models appear appropriate for remote self‐administered measures and are sensitive to cognitive impairment. Device type and response input did not explain enough variance to warrant inclusion in normative models but are important for individual level interpretation. Future work will improve representation of individuals from under‐represented groups.

AbstractBackgroundLimbic‐predominant age‐related TDP‐43 encephalopathy (LATE) is a neuropathologically‐defined disease, and it is frequently comorbid with Alzheimer’s disease neuropathological change (ADNC). However, the neurological syndrome associated with LATE neuropathological change (LATE‐NC) is not defined. We propose a set of clinical criteria for a limbic‐predominant amnestic neurodegenerative syndrome (LANS) that is highly associated with LATE‐NC. These criteria assess degree of certainty (highest, high, moderate, low) based on features that are measurable in vivo, including older age at evaluation, a mild clinical syndrome, impaired semantic knowledge, disproportionate hippocampal atrophy, limbic hypometabolism, absence of neocortical degeneration patterns and low likelihood of neocortical tau. We operationalized these criteria using two large datasets with clinical and pathologic outcomes.MethodWe screened autopsied patients from Mayo Clinic and ADNI cohorts and applied the LANS criteria to those with an antemortem predominant amnestic syndrome (Mayo, n = 165; ADNI, n = 53). We compared frequencies of neuropathological diagnoses across LANS likelihoods, compared longitudinal CDR‐SB trajectories across LANS likelihoods, and stratified ADNC/LATE‐NC patients according to their LANS likelihood and compared their longitudinal CDR‐SB trajectories to those with ADNC or LATE‐NC.ResultADNC, ADNC/LATE‐NC and LATE‐NC accounted for 35%, 37% and 4% of cases in the Mayo cohort, respectively, and 30%, 22%, and 9% of cases in the ADNI cohort, respectively (Fig. 1). ADNC cases had the lowest LANS likelihoods, LATE‐NC patients had the highest likelihoods, and ADNC/LATE‐NC patients had intermediate likelihoods. Patients with high LANS likelihoods had a milder and slower clinical course and more severe temporo‐limbic degeneration compared to those with low likelihoods (Fig. 2). ADNC/LATE‐NC patients with higher likelihoods had more temporo‐limbic degeneration and a slower rate of cognitive decline, and those with lower likelihoods had more lateral temporo‐parietal degeneration and a faster rate of cognitive decline (Fig. 3).ConclusionThe implementation of LANS criteria has implications for disambiguating the different driving etiologies of progressive amnestic presentations in older age to guide prognosis, treatment, and clinical trial design and enrollment. The development of in vivo biomarkers specific to TDP‐43 pathology are needed to further refine molecular associations between LANS and LATE‐NC and precise antemortem diagnoses of LATE.