Toronto Dementia Research Alliance

BackgroundThe long‐term consequences of repetitive head impacts in contact sports include cognitive deficits, accelerated brain atrophy, and neurodegenerative diseases. Currently, no studies of former athletes have addressed the connection between brain aging and biological aging, which can be assessed using age‐related DNA methylation (DNAm) profiles. The most studied epigenetic clock is DNAm‐age, which is providing consistent results across tissues (e.g., blood and brain). It includes several measures like DNAm‐age acceleration (DNAmAA), the difference between DNAm‐age and chronological age, and DNAmAA‐residual (DNAmAAr), independent of chronological age. In a cohort of retired athletes, we explored the link between measures of brain age and epigenetic age, including the recently developed DNAmFit‐age reported to be younger in physically fit individuals.MethodWe investigated 126 former contact sports athletes (mean age: 54.5±14.4; 96% male; mean concussion number: 6.8±6.7). Longitudinal assessments were available for 21 athletes (2–3 time points over 1–10 years). Bisulfite‐converted blood DNA was analyzed using the Infinium MethylationEPIC chip, and the DNAm data were submitted to the Horvath calculator (https://dnamage.clockfoundation.org/) to obtain DNAmFit‐age, DNAm‐age, and DNAmAAr. T1‐weighted MRIs were processed using the CAT12‐Toolbox. Regional gray matter volume was examined in relation to DNAmAA, DNAmAAr and DNAmFit‐age acceleration (DNAmFitAA).ResultConsistent with previous findings, chronological age was significantly associated with brain volumes and cortical thickness. DNAmAA, DNAmAAr, and DNAmFitAA remained stable over a period of up to 10 years, and none of these measures were associated with brain volumes or cortical thickness. Notably, multivariate linear regression analysis revealed a significant positive association between DNAmFitAA and the number of concussions (p = 0.0027, B=0.46, R2=0.063), indicating that every additional two concussions correspond to a 5‐year increase in DNAmFitAA.ConclusionIn former athletes, we confirmed that chronological age is associated with cerebral atrophy and identified an association between increased DNAmFitAA and a higher number of concussions. This finding requires validation in independent studies, along with an assessment of the relationship between DNAmFitAA and neurodegeneration. Our analysis did not reveal a link of the examined epigenetic clocks with brain volumes and cortical thickness. Exploration of other epigenetic clocks may provide new insights into the mechanisms underlying brain aging.

BackgroundThere is substantial heterogeneity in clinical presentation of genetic Frontotemporal Dementia (FTD), even within the same family. This suggests that additional heritability may exist and contribute to this variable presentation. We examined whether gene‐based aggregate burden of genome‐wide rare variants (minor allele frequency [MAF]: ≤1%) contribute to variation in regional cortical and subcortical grey matter volumes, after controlling for effects of causative mutations in GRN, MAPT, and C9orf72.MethodThis study was embedded within the GENetic Frontotemporal dementia Initiative (GENFI), which recruits genetic FTD cases and their asymptomatic at‐risk family members, both carriers and non‐carriers of FTD mutations. We included 518 participants with genotype (Neurochip; imputed against TOPMed), and T1w‐MRI brain volumetric data. Gene‐based burden tests that aggregate the number of rare variants by gene were used to examine the association of rare variants (MAF: ≤1%) with regional cortical and subcortical grey matter volumes (70 regions of interest [ROIs]), controlling for age, sex, total intracranial volume, mutation status, scanner site, population stratification, and family membership (kinship matrix) using RVTests. Annotations for loss of function mutations (LOF): start gain, stop loss, start loss, essential splice site, stop gain, normal splice site, and non‐synonymous. Multiple testing correction accounted for the number of genes and number of independent grey matter volumes as calculated by matSpD (p‐value threshold: 0.05/(17,053x42) = 6.98 x10‐8).ResultAggregate burden of LOF mutations (DNAJB8‐AS1, WDR26, RDM1P5, BSND, CNOT2, DDA1, ASAH2B, PPM1A, HOXD13, ALDH1A1, CENATAC, ANKRD45) was associated with significantly lower volumes within the left temporal lobe (ROIs: left temporal and lateral temporal left), and greater volume in the putamen bilaterally (TSACC). All genes are protein coding, except the DNAJB8‐AS1 (antisense RNA) and RDM1P5 (pseudogene), and are variably expressed in the brain. Molecular functions of significant genes involve regulation of gene expression, transcription, and cell cycle, ion channel function, and chromosomal segregation. WDR26 and CNOT2 genes have been implicated in neurodevelopment and neurological disorders respectively; BSND gene is involved in neurotransmission.ConclusionIdentification of deleterious or protective rare variants contributing to FTD imaging phenotypes may help identify genetic modifiers of familial FTD. Replication in larger cohorts is needed.

BackgroundAmnestic mild cognitive impairment (aMCI) is characterized by synaptic loss and cognitive decline and is considered a precursor to Alzheimer’s Disease. Currently, there are no effective treatments for aMCI, and available treatments (i.e., cholinesterase inhibitors) do not appear to improve cognitive or functional outcomes. Psychedelics, including psilocybin, have regained interest for treatment of treatment‐resistant neuropsychiatric disorders. Research suggests psilocybin’s psychedelic and clinical effects may be due to interaction with the 5HT2A serotonin receptor (5HTA‐R) in the brain. Cognitive impairments, such as memory decline, have been associated with lower 5HT2A‐R density in the brain. Encouragingly, preclinical animal studies suggest that psilocybin may promote synaptogenesis in the brain, particularly in areas associated with learning and memory, likely through its interaction with the 5HT2A‐R. Psilocybin may represent a novel treatment to counter neurodegenerative progression and consequently improve cognitive outcomes in patients with aMCI.MethodThe present double‐blind, placebo‐controlled randomized PET study will use the radioligand [18F]SynVesT to assess psilocybin’s effect on synaptic density in the hippocampus and prefrontal cortex of patients with aMCI, and whether these changes are associated with improved cognitive outcomes. aMCI participants and sex‐matched healthy controls will be randomized to receive either two doses of 25mg psilocybin or placebo one week apart. Participants will be monitored by a study physician and qualified therapist. PET scans are conducted pre‐ and one‐week post‐treatment. Clinical, safety, and neuropsychological assessments are done at baseline and 1‐, 4‐, and 12‐weeks post‐treatment. Safety measures include monitoring vital signs, suicidal ideation, and adverse events (AEs).ResultPilot data from two aMCI participants (Male% = 50%, mean age = 71.0(±3.0), mean baseline MOCA = 22(±3.0)) and three healthy controls (Male% = 67%, mean age = 66.3(±5.1); mean baseline MOCA = 27.7(±1.5)) is available. All participants completed study procedures without issue. Psilocybin was well tolerated, with no unexpected or serious AEs. All expected AEs resolved without sequelae. Expected AEs included dizziness (n=4) and altered perception (n=3). Blinding remains in effect.ConclusionThe preliminary data suggest psilocybin is safe, well‐tolerated, and can be feasibly investigated in a supervised medical setting in older adults as a prospective treatment for aMCI.

BackgroundAccurate clinical diagnosis for Alzheimer's disease dementia (AD), amnestic mild cognitive impairment (aMCI), and non‐amnestic MCI (naMCI) is essential for timely management. The diagnosis is made using a range of factors including cognitive testing. Explainable artificial intelligence (XAI)‐based SHAP (SHapley Additive exPlanations) is a machine learning interpretability tool that can provide insights into specific features that drive classification decisions. We used XAI with support vector machines (SVM) to identify key cognitive features of Toronto Cognitive Assessment (TorCA), a user‐friendly cognitive assessment administered by frontline clinicians, for differentiating neurocognitive disorder diagnoses.MethodWe used data from the Toronto Dementia Research Alliance (TDRA) database, comprising of participants with AD, aMCI, naMCI, or normal cognition (NC) seen in memory clinics across Toronto. An SVM model with radial basis function (RBF) kernel was configured with 10‐fold cross‐validation. XAI was integrated using SHAP values to identify the most important critical features contributing to the model predictions. Classification accuracies, defined as the proportion of correct classifications for each pairwise comparison, were calculated using TorCA total scores and specific features from subtests.ResultWe included 695 participants (149 AD, 189 aMCI, 304 naMCI, and 53 NC). Classification accuracy for distinguishing AD vs NC was excellent, whether using all TorCA subtests (0.97±0.03), or the top 5 features (0.96±0.04) (Delayed Recall, Immediate Recall Trials 1 and 2, Sentence Comprehension, Benson Figure Recall), but lower (0.93±0.06) with only TorCA total score. Classification accuracy for MCI or naMCI vs NC was also very good (0.86±0.03 to 0.89±0.04) using the top 5 features.ConclusionTorCA combined with XAI can accurately differentiate common clinical neurocognitive disorder phenotypes in ambulatory settings. These findings point to specific cognitive subtests important for diagnosis and may help improve the efficiency of cognitive testing. Future studies should investigate the differentiation of other neurocognitive disorders using these tools and further validate these findings using formal neuropsychological testing.

BackgroundSubjective cognitive decline (SCD), defined as self‐perceived memory decline with normal test performance, may indicate preclinical Alzheimer’s disease (AD). While medial temporal lobe (MTL) atrophy is a key neuroimaging biomarker for AD, specific atrophy patterns in SCD remain unclear. Neuroimaging studies also indicate that hippocampal subfields, particularly the CA1 and subiculum, are among the earliest affected in AD. However, the literature on volumetric alterations in the hippocampal subfields and other MTL subregions in SCD is inconsistent. One potential reason is failure to account for previous diagnosis of depression, which itself is associated with hippocampal atrophy, in SCD samples. This study compared MTL and hippocampal subfield volumes in older adults with SCD and cognitively unimpaired (CU) adults without prior or current depression and examined their associations with quantitative measures of subjective memory.MethodFifty‐one participants (26 SCD (13M, age 70.6 ± 5.1) and 25 CU (10M, age 71.4 ± 7.4) without psychiatric history completed the Memory Functioning Questionnaire (MFQ) and high‐resolution magnetic resonance imaging (MRI) of the MTL at 3Tesla. The Automatic Segmentation of Hippocampal Subfields (ASHS) pipeline and manual segmentation were used to extract regions‐of‐interest (ROIs). ANCOVA was used to compare group differences in ROIs. Regression analyses were conducted to examine associations between MFQ scores and ROIs.ResultRelative to CU, parahippocampal volume was smaller bilaterally in SCD [left: F(1, 48 =5.87, p =0.019; right: F (1, 48)=5.35, p =0.026], after controlling for age, sex, and total intracranial volume. A trend towards smaller left subicular volume was observed in SCD [F(1,48)=3.86, p =0.055]. Multivariable regression analyses controlling for age, sex, and intracranial volume showed positive associations between left subiculum volumes and MFQ‐Seriousness of Forgetting subscale scores, and between left CA1 volumes and MFQ‐Frequency of Forgetting subscale scores.ConclusionThese preliminary findings implicate the subiculum and parahippocampal gyrus in the neurobiology of SCD and suggest a predominant effect of the left hemisphere. Further, these data are not confounded by the presence of past or current depression in our sample. These findings require replication using larger samples of SCD with biomarker evidence of AD neuropathology.

Context: The systematic evaluation of a practitioner’s adherence to and competence in delivering psychotherapeutic interventions can be complex. This study describes the development of a fidelity assessment tool for the Reitman Centre CARERS Program (RCCP), a carer group-psychotherapy intervention with multiple didactic and clinical components. The tool’s value in informing psychotherapy training and best practices for practitioners from diverse professional settings is examined. Methods: The RCCP Fidelity Assessment Tool (RCCP-FAT) was developed following an iterative process of item writing and checking. Seven components of the RCCP—Group Structure, Dementia Education, Problem-Solving Techniques, Therapeutic Simulation, Vertical Cohesion, Horizontal Cohesion, and Overall Global Rating—were assessed, with three to eight items, and a “global score” assigned to each. Fifteen trained raters were paired up to rate 36 RCCP sessions using the RCCP-FAT. Rater agreement, correlation between itemized and global scores, and correlation between global ratings and RCCP participants’ satisfaction were calculated. Results: A total of 1188 RCCP-FAT items were rated by each of the two rater groups. Rater agreement was calculated to be 54.3% (κ = 0.32; 95% CI, 0.02681–0.3729). A positive correlation was found between the itemized and global scoring for four RCCP components evaluated (R = 0.833 to 0.929; p < 0.01). The global score and the participants’ satisfaction with “Simulation” was also positively correlated (R = 0.626, p < 0.01). Conclusions: The study provided evidence for fair rater agreement for all RCCP-FAT assessment items. More importantly, the process of developing the tool systematically crystallized the clinical elements of the RCCP and helped to standardize the training methods by creating a framework for providing feedback to learners that matches the items on the RCCP-FAT. The use of the RCCP-FAT to guide the training and mentoring of incoming group leaders is essential in the scaling and dissemination of a complex training method like the RCCP to ensure fidelity to the original evidence-based intervention.

Abstract BackgroundWhite matter hyperintensities (WMH) are commonly observed on MRI in Alzheimer’s disease (AD), but the molecular pathways underlying their relationships with the ATN biomarkers remain unclear. The aim of this study was to identify genetic variants that may modify the relationship between WMH and the ATN biomarkers.MethodThis genome‐wide interaction study (GWIS) included individuals with AD, MCI, and normal cognition from ADNI (n = 1012). WMH were measured from FLAIR MRI using an automated atlas‐based segmentation. CSF Aβ42, tau, and p‐tau concentrations were measured using Roche Elecsys immunoassays. Hippocampal volumes were obtained using FreeSurfer. Interaction effects between single nucleotide polymorphisms (SNP) and WMH volumes on ATN biomarkers were assessed using linear regression models, adjusting for age, sex, diagnosis, MMSE, APOE‐ε4 status, head‐size, and 4 genetic principal components in PLINK2. Linkage disequilibrium‐based clumping (LD: 0.5; physical distance: 250kb) was performed in PLINK1.9, visualized with LocusZoom. Significant SNP‐WMH interactions discovered in ADNI were validated in participants from the Sunnybrook Dementia Study (n = 433) and UK Biobank (n = 16720).ResultA 30‐SNP intergenic locus on chromosome 18 (lowest p‐value‐SNP: rs72899960 T>A, β = 227.0 pg/ml, SE = 10.2, p = 2.30 × 10‐8, MAF = 11.1%, Imputation‐R2>0.99%, n = 848) achieved genome‐wide significance interacting with WMH to predict Aβ42, with the nearest gene being a novel long noncoding RNA (lncRNA), ENSG00000286844 (‐360KB). A 9‐SNP intronic locus on chromosome 11 (lowest p‐value‐SNP: rs3912008 C>T, β = 0.25 cm3, SE = 0.044, p = 1.47 × 10‐8, MAF = 22.6%, Imputation‐R2>99%, n = 987) achieved genome‐wide significance interacting with WMH to predict hippocampal volume, and was found within the microRNA gene MIR4300HG. Both SNPs showed additive moderation effects such that associations between WMH and ATN markers in major allele homozygotes were diminished in heterozygotes, and diminished further in minor allele homozygotes.The interaction between MIR4300HG‐rs3912008 and WMH in predicting hippocampal volume replicated in dominant models in both the Sunnybrook Dementia Study (F1,431 = 7.98, coefficient = 0.14 cm3, SE = 0.05, p = 0.005) and in UK Biobank (F1,16623 = 5.32, coefficient = 0.025 cm3, SE = 0.011, p = 0.021).ConclusionGenomic variants moderate relationships between WMH and ATN biomarkers suggesting a novel regulatory role for a long non‐coding RNA, and the involvement of microRNA MIR4300HG. This interaction study may help identify the contribution of specific genes and the potential pathways involved in the relationships between WMH and AD biomarkers.

Older adults with major depressive disorder (MDD) or mild cognitive impairment (MCI) are at high risk for cognitive decline. To assess the efficacy of cognitive remediation (CR) plus transcranial direct current stimulation (tDCS) targeting the prefrontal cortex in slowing cognitive decline, acutely improving cognition, and reducing progression to MCI or dementia in older adults with remitted MDD (rMDD), MCI, or both. This randomized clinical trial was conducted at 5 academic hospitals in Toronto, Ontario, Canada. Participants were older adults who had rMDD (with or without MCI, age ≥65 y) or MCI without rMDD (age ≥60 y). Assessments were made at baseline, month 2, and yearly from baseline for 3 to 7 years. CR plus tDCS (hereafter, active) or sham plus sham 5 days a week for 8 weeks followed by twice-a-year 5-day boosters and daily at-home CR or sham CR. The primary outcome was change in global composite cognitive score. Secondary outcomes included changes in 6 cognitive domains, moderating effect of the diagnosis, moderating effect of APOE ε4 status, change in composite score at month 2, and progression to MCI or dementia over time. Of 486 older adults who provided consent, 375 (with rMDD, MCI, or both) received at least 1 intervention session (mean [SD] age, 72.2 [6.4] years; 232 women [62%] and 143 men [38%]). Over a median follow-up of 48.3 months (range, 2.1-85.9), CR and tDCS slowed cognitive decline in older adults with rMDD or MCI (adjusted z score difference [active - sham] at month 60, 0.21; 95% CI, 0.07 to 0.35; likelihood ratio test [LRT] P = .006). In the preplanned primary analysis, CR and tDCS did not improve cognition acutely (adjusted z score difference [active - sham] at month 2, 0.06, 95% CI, -0.006 to 0.12). Similarly, the effect of CR and tDCS on delaying progression from normal cognition to MCI or MCI to dementia was weak and not significant (hazard ratio, 0.66; 95% CI, 0.40 to 1.08; P = .10). Preplanned analyses showed treatment effects for executive function (LRT P = .04) and verbal memory (LRT P = .02) and interactions with diagnosis (P = .01) and APOE ε4 (P < .001) demonstrating a larger effect among those with rMDD and in noncarriers of APOE ε4. The study showed that CR and tDCS, both targeting the prefrontal cortex, is efficacious in slowing cognitive decline in older adults at risk of cognitive decline, particularly those with rMDD (with or without MCI) and in those at low genetic risk for Alzheimer disease. ClinicalTrials.gov Identifier: NCT02386670.

Background: While open datasets are adopting FAIR principles to improve their discovery and use, restricted data—those only accessible via request or application—have fallen behind. Metadata is not an inherent characteristic of restricted data, which limits its ability to be found and used. To better understand discoverability and accessibility of restricted data, this study reviewed restricted health data sources to determine how they describe their datasets and access procedures, what descriptive commonalities exist across data sources, and to what extent the commonalities we found can be accommodated within existing metadata schemas. Methods: This study extracted dataset and access information provided by a sample of 48 restricted data sources, identified commonalities across these data sources to develop possible metadata elements for restricted data, and mapped these metadata elements to existing metadata schemas (e.g., DataCite) to evaluate how well they accommodate information supplied by restricted data sources. Results: Restricted data sources describe their datasets (35 commonalities) and access procedures (27 commonalities) in similar ways. Dataset descriptions aligned with existing metadata schemas, with the DDI-Lifecycle and -Codebook schemas receiving 91.4% and 85.7% exact matches respectively with the dataset elements we identified. Access procedures did not align with metadata available in existing schemas. Discussion: While descriptive dataset metadata for restricted data sources will make their data more findable, the accessibility of these datasets could be significantly improved by structured metadata capturing data access information. Presently, metadata schemas do not accommodate the level of detail restricted data sources provide about access procedures and requirements.