Citations

Abstract Grey matter network topology is altered in Alzheimer’s disease and these alterations are related to cognitive decline. Understanding the biological underpinnings of loss of brain connectivity may provide insights into mechanisms related to developing Alzheimer’s dementia (i.e. dementia A+). We investigated which biological processes as measured in CSF proteomics were associated with loss of brain connections across the Alzheimer’s disease continuum. We included 347 individuals with abnormal CSF amyloid [mean age ± standard deviation (SD) 66 ± 8; 98 cognitively unimpaired—A+, 88 mild cognitive impairment—A+, 161 dementia A+] and 146 cognitively unimpaired individuals with normal CSF amyloid (mean age ± SD 62 ± 8) and available T1w MRI-scans and CSF proteomic data (3097 proteins using tandem mass tag spectrometry) from the Amsterdam Dementia Cohort. We used an automated pipeline to construct grey matter networks from 3D-T1 sequences and for each network, calculated the small-worldness coefficient, which we previously found to be robustly related to cognitive decline. Linear models were applied to test associations between CSF protein levels and connectivity measures using an interaction term for clinical stage while controlling for connectivity density, age and sex. We validated our results in data from the Alzheimer’s disease Neuroimaging Initiative (ADNI). Pathway enrichment analysis was performed for proteins associated with loss of brain connectivity (P < 0.05) using the Gene Ontology database. Individuals across the Alzheimer’s disease continuum had lower small-worldness coefficients compared with controls (ANOVA P < 0.001). In amyloid positive individuals, higher levels of 222 proteins and lower levels of 482 proteins were associated with lower small-worldness coefficients and were enriched for innate immune system and neuroplasticity pathways, respectively. Stratified for disease stage, most protein associations with lower small-worldness coefficients were found in mild cognitive impairment A+ (n = 527 proteins) and dementia A+ (n = 799 proteins) with considerable overlap (n = 239 proteins). Proteins in these stages were enriched for complement activation and synaptic integrity. In cognitive unimpairment A+, we found proteins enriched for processes involved in apoptosis. We did not find any enriched biological processes in controls. Repeating analyses in ADNI indicated that similar biological processes were associated with altered grey matter network connectivity. Higher CSF levels of proteins involved in immune responses and lower levels of proteins related to neuroplasticity were associated with lower small-worldness coefficients across the Alzheimer’s disease continuum. This suggests that preserving cognitive function in the presence of amyloid and prevention of dementia A+ may require therapies that strengthen synapses and targets the innate immune system in addition to amyloid and tau.

Abstract INTRODUCTION The Centiloid scale is the standard for amyloid (Aβ) PET quantification in research and clinical settings. However, variability between tracers and scanners remains a challenge. METHODS This study introduces DeepSUVR, a deep learning method to correct Centiloid quantification, by penalizing implausible longitudinal trajectories during training. The model was trained using data from 2,129 participants (7,149 Aβ positron emission tomography [PET] scans) in the Australian Imaging, Biomarkers and Lifestyle Study of ageing (AIBL)/Alzheimer's Disease Neuroimaging Initiative (ADNI) and validated using 15,807 Aβ PET scans from 10,543 participants across 10 external datasets. RESULTS DeepSUVR increased correlation between tracers, and reduced variability in the Aß‐negatives. It showed significantly stronger association with cognition, visual reads, neuropathology, and increased longitudinal consistency between studies. DeepSUVR also increased the effect size for detecting small treatment related slowing of amyloid accumulation in the A4 study. DISCUSSION DeepSUVR substantially advances Aβ PET quantification, outperforming all standard approaches, which is particularly important for clinical decision making and to detect subtle or early changes in Aβ. Highlights Novel artificial intelligence (AI)‐method that penalizes biologically implausible longitudinal trajectories, enabling the model to learn standardized uptake value ratios (SUVR) correction factors without requiring longitudinal data at inference time. Improves Centiloid consistency across tracers and studies, significantly enhancing cross‐sectional and longitudinal amyloid positron emission tomography (PET) quantification. DeepSUVR‐derived Centiloids show stronger associations with cognition, visual reads, and neuropathology. Longitudinal variability is reduced three‐ to five‐fold, enabling more reliable tracking of amyloid accumulation and better detection of treatment effects. Novel reference and target masks derived from DeepSUVR replicate most of the model's performance, offering a practical alternative for integration into existing pipelines.

Abstract Background Cerebrospinal fluid (CSF) seed amplification assays (SAAs) for detecting α‐synuclein (αsyn) seeds have recently emerged as robust in vivo biomarkers of Lewy body pathology (LBP). However, CSF sampling is invasive, costly, and time‐consuming, limiting its widespread use. We aimed to develop and evaluate a neuropathologically‐validated two‐step workflow, leveraging smell function and CSF αsyn SAA to accurately determine postmortem LBP status. This approach could reduce the number of confirmatory lumbar punctures needed (Figure 1). Methods The study included 358 individuals from the Arizona Study of Aging and Neurodegenerative Disorders with antemortem smell testing, CSF αsyn SAA results, and postmortem neuropathological assessments of regional LBP burden. Step‐1 of the two‐step workflow involved a risk‐stratification model predicting postmortem cortical LBP‐positivity (LBP ctx , defined as at least mild pathology in any cortical region) using logistic regression models with University of Pennsylvania Smell Identification Test (UPSIT) scores (smell testing), age, and sex as predictors. In step‐2, confirmatory CSF αsyn SAA testing was applied only to participants identified as high‐risk in step‐1. Workflow performance ‐ accuracy, positive predictive value (PPV), negative predictive value (NPV), and reduction in CSF testing ‐ was evaluated in (i) the entire study cohort; (ii) patients with clinical parkinsonism; (iii) patients with an Alzheimer's disease (AD) clinical syndrome; and (iv) clinically unimpaired (CU) individuals. Results Participants had a mean age at death of 86.2±7.8 years, 42.6% were female, 35.2% had LBP ctx , and the average time between UPSIT and death was 3.2±2.3 years (Table 1). Using a 95% sensitivity cut‐off for the UPSIT‐based algorithm, the two‐step workflow achieved high accuracy in identifying LBP ctx (whole cohort=94%; clinical parkinsonism=95%; clinical AD=94%; CU=93%; Figure 2a), while reducing CSF testing (whole cohort=‐43%; clinical parkinsonism=‐23%; clinical AD=‐35%; and CU=‐80%; Figure 2d). PPVs ranged from 75‐96% and were highest in the clinical parkinsonism subgroup (96%) where LBP ctx + was highest (62.7%). NPVs ranged between 95‐98% and were highest in the clinical AD subgroup (98%) (Figure 2b‐c). The two‐step approach reached accuracies similar to using CSF tests in all participants. Conclusions Implementing a two‐step workflow in different clinical scenarios may reduce invasive testing with CSF, minimizing the burden for individuals and costs for healthcare providers.

Abstract The Centiloid scale is the standard for Amyloid (Aβ) PET quantification, widely used in research, clinical settings, and trial stratification. However, variability between tracers and scanners remains a challenge. This study introduces DeepSUVR, a deep learning method to correct Centiloid quantification, by penalising implausible longitudinal trajectories during training. The model was trained using data from 2,098 participants (6,762 Aβ PET scans) in AIBL/ADNI and validated using 15,806 Aβ PET scans from 10,543 participants across 10 external datasets. DeepSUVR increased correlation between tracers, and reduced variability in the Aβ-negatives. It showed the strongest association with cognition, highest AUC against visual reads and best longitudinal consistency between studies. DeepSUVR also increased the effect size for detecting lower Centiloid increase per year in the A4 study. DeepSUVR advances Aβ PET quantification, outperforming standard approaches, which is particularly important for consistent decision making and to detect subtle and early changes in clinical interventions.

The translation of AI-generated brain metastases (BM) segmentation into clinical practice relies heavily on diverse, high-quality annotated medical imaging datasets. The BraTS-METS 2023 challenge has gained momentum for testing and benchmarking algorithms using rigorously annotated internationally compiled real-world datasets. This study presents the results of the segmentation challenge and characterizes the challenging cases that impacted the performance of the winning algorithms. Untreated brain metastases on standard anatomic MRI sequences (T1, T2, FLAIR, T1PG) from eight contributed international datasets were annotated in stepwise method: published UNET algorithms, student, neuroradiologist, final approver neuroradiologist. Segmentations were ranked based on lesion-wise Dice and Hausdorff distance (HD95) scores. False positives (FP) and false negatives (FN) were rigorously penalized, receiving a score of 0 for Dice and a fixed penalty of 374 for HD95. The mean scores for the teams were calculated. Eight datasets comprising 1303 studies were annotated, with 402 studies (3076 lesions) released on Synapse as publicly available datasets to challenge competitors. Additionally, 31 studies (139 lesions) were held out for validation, and 59 studies (218 lesions) were used for testing. Segmentation accuracy was measured as rank across subjects, with the winning team achieving a LesionWise mean score of 7.9. The Dice score for the winning team was 0.65 ± 0.25. Common errors among the leading teams included false negatives for small lesions and misregistration of masks in space. The Dice scores and lesion detection rates of all algorithms diminished with decreasing tumor size, particularly for tumors smaller than 100 mm3. In conclusion, algorithms for BM segmentation require further refinement to balance high sensitivity in lesion detection with the minimization of false positives and negatives. The BraTS-METS 2023 challenge successfully curated well- annotated, diverse datasets and identified common errors, facilitating the translation of BM segmentation across varied environments and providing the tools for future development of personalized volumetric reports to patients undergoing BM treatment.

AbstractBackgroundRecent advances in automatic face recognition have increased the risk that de‐identified research imaging data could be re‐identified from face imagery in brain scans.MethodAn ADNI committee of independent imaging experts evaluated 11 published techniques for face‐deidentification (“de‐facing”) and selected four algorithms (FSL‐UK Biobank, HCP/XNAT, mri_reface, and BIC) for formal testing using 183 longitudinal scans of 61 racially and ethnically diverse ADNI participants, evaluated by their facial feature removal on 3D rendered surfaces (confirming sufficient privacy protection) and by comparing measurements from ADNI routine image analyses on unmodified vs. de‐faced images (confirming negligible side effects on analyses).ResultEffects of the de‐facing methods on FreeSurfer at UCSF (Figure 1‐top), ASHS at University of Pennsylvania (Figure 1 bottom‐left), and longitudinal atrophy at Mayo Clinic (Figure 1 bottom‐right) were comparable with each other and sufficiently small vs. unmodified scans to recommend de‐facing in ADNI‐4. The FSL and mri_reface methods were more thorough in removing facial features, partly because these also removed ears (Figure 2‐top). mri_reface was recommended for ADNI‐4 primarily because fewer (zero) images had modified brain (Figure 2‐bottom), and ADNI leadership approved de‐facing with mri_reface. ADNI performed a second validation of its latest version on another diverse sample of 100 ADNI‐3 participants. These analyses included: boundary shift interval at University College London, where average differences in measurements were all <0.05%; cortical white matter hyperintensities at UC Davis, where they found “minimum impact” (R>0.97); and others.ConclusionADNI‐4 de‐faces all indicated image types (Table 1) before subsequent pre‐processing, analyses, and public release. Sites upload images to the central repository (LONI), where they are queued to the MRI Core at Mayo Clinic for de‐facing. Image types requiring de‐facing are de‐faced using mri_reface. Trained analysts inspect de‐faced images to confirm complete removal of the face and complete non‐modification of the brain, and any flagged images are re‐processed manually until issues are resolved. De‐faced images are converted back to DICOM and re‐uploaded for further distribution. Image types not requiring de‐facing are indicated to proceed directly to subsequent steps. In summary, ADNI‐4 has adopted mri_reface because it successfully de‐faces without substantially affecting brain measurements.

AbstractBackgroundPhase four of the Alzheimer’s Disease Neuroimaging Initiative (ADNI4) began in 2023. This time‐period corresponded to MRI vendors introducing product sequences with compressed sensing (CS), cross‐vendor adoption of arterial spin‐labelling (ASL) and multi‐band slice excitation, and hardware improvements (head‐coils, increased gradient amplitudes). These advances enabled the acquisition of new imaging measures and reduced scan times. The ADNI4 MRI protocol aims to maintain longitudinal consistency across two decades of data acquisition, while adopting new technologies. Here we describe the design and justification behind the ADNI4 protocol and list the target imaging biomarkers.MethodsThe ADNI4 MRI protocol includes 9 MRI sequences; i) T1‐weighted (T1w) MP‐RAGE, ii) a compressed sensing T1w MP‐RAGE, iii) 3D T2w FLAIR, iv) structural isotropic 3D T2w v) T2*‐weighted susceptibility imaging vi) multi‐post labelling delay (PLD) 3D ASL, vii) 2D multi‐band‐multiple‐shell diffusion (dMRI), viii) 2D multi‐band EPI BOLD ix) high resolution, limited field of view, imaging centered on the hippocampus. Table 1 gives a detailed summary of the key protocol parameters and lists some of the imaging biomarkers that can be obtained from each series.ResultsStructural imaging was kept consistent across imaging sites. Variations in vendor specific sequences, software upgrades, and hardware performance capabilities, resulted in variations in advanced imaging strategies. Table 2 highlights the advanced sequences introduced in ADNI4, which will gradually be implemented as the required upgrades and licenses are obtained at each site. The main advanced sequences that are new in ADNI4 from ADNI 3 are i) CS for T1w images, ii) pseudo‐continuous‐ASL (pCASL) being implemented on all 3 vendors (GE, Siemens, Philips), iii) multi‐PLD ASL, which allows for the generation of quantitative maps of cerebral blood‐flow (CBF) and arterial transit times (ATTs), iv) and isotropic 1mm3 resolution for 3D FLAIR. When ADNI4 began, product sequences that utilized artificial intelligence were not widely available and could not be adequately tested before the rollout of ADNI4.ConclusionsThe ADNI4 database will continue to help the neuroimaging community, not only extract valuable imaging biomarkers, but provide the data necessary to test the impact of advanced imaging strategies on diagnostic accuracy and disease sensitivity.

AbstractBackgroundScalable, efficient methods are needed to enroll diverse populations of older adults into AD observational studies and clinical trials. We evaluated preliminary feasibility of a novel, digital, culturally informed approach to recruit and screen participants for the Alzheimer’s Disease Neuroimaging Initiative (ADNI4).MethodsDigital advertising tailored towards Black/African American and Latinx older adults residing near six clinical ADNI sites directed potential participants to a recruitment website (Figure 1). They completed digital surveys of demographics, ADNI exclusion criteria, memory concerns and changes, self‐report cognitive impairment, and the Everyday Cognition Scale (ECog)‐12 item. Participants also completed Novoic Storyteller, a self‐administered speech‐based cognitive test. Digital assessment results were used to prioritize those from minoritized ethnocultural groups, with low educational attainment, and possible Mild Cognitive Impairment (MCI) for remote blood draw at local Quest phlebotomy centers to obtain AD plasma biomarkers. Completion rates and digital assessment performance were compared between minoritized and non‐minoritized groups.ResultsOf 989 participants recruited from digital advertising efforts who provided contact information, 230 (23%) signed consent, and 183 (19%) completed at least one remote assessment (Table 1). 45% were from minoritized ethnocultural groups, 4% had ≤12 years of education, and 53% had evidence of possible MCI from digital assessments. 34 (19%) were referred to the plasma biomarker study, of which 21% completed blood draw. Compared to non‐minoritized participants, minoritized participants were less likely to report a diagnosed cognitive impairment or a family history of AD, less likely to have ECog scores indicative of MCI, and less likely to have a study partner. Minoritized participants had lower ECog and Novoic Storyteller completion rates.ConclusionRecruitment and assessment of a diverse cohort of older adults, including those with possible cognitive impairment, is feasible using culturally informed digital advertising. Improving study engagement and achieving educational diversity are key challenges. This approach is now being scaled up to facilitate recruitment into in‐clinic ADNI4, with the goal of enrolling 500 new participants: >50% from minoritized groups, 40% with MCI, and 80% amyloid positive across diagnostic groups. This approach can be adapted to facilitate recruitment and longitudinal assessment in other AD studies and trials.

AbstractBackgroundRecent studies highlight distinct patterns of cortical atrophy between amnestic (typical) and non‐amnestic (atypical, with subtypes: behavioural, dysexecutive, logopenic and visuospatial) clinical phenotypes of Alzheimer’s disease (AD). The current study aimed to assess regional MRI patterns of cortical atrophy across AD phenotypes, and their association with amyloid‐beta (Aß), phosphorylated tau (pTau), axonal degeneration (NfL) and microvascular deterioration (COLIV).MethodPostmortem In‐situ 3DT1 3T‐MRI data was collected for 33 AD (17 typical, 16 atypical) and 16 control brain donors. Images were segmented and AAL3 atlas regional volumes were obtained using QyScore®. At subsequent autopsy, eight brain regions were selected, immunostained for Aß (4G8), pTau (AT8), Neurofilament‐light (NFL), and Collagen‐IV (COLIV), and quantified using qupath. Group comparisons and volume‐pathology associations were analyzed using linear models and partial correlations with covariates age, sex, postmortem delay, and intracranial volume.ResultsCompared to controls, AD phenotype groups showed overall lower cortical volume, while only minor volume differences were observed between AD phenotype groups, observed primarily in limbic regions (Figure 1). Across pathological markers, AD phenotype groups showed consistently higher immunoreactivity than controls, while atypical AD showed consistently higher immunoreactivity than typical AD (Figure 2). Moreover, different patterns of pathology could be observed between atypical subtypes (e.g. distinctly higher pTau load in the occipital gyrus of the visuospatial subtype). In typical AD, global volume loss was associated with lower Aß and higher pTau, NFL and COLIV immunoreactivity, while in atypical AD, global volume loss was primarily associated with higher NFL immunoreactivity (Figure 3a). Regionally, AD phenotype differences in atrophy‐pathology association were most pronounced in the (para)hippocampal regions. This distinction was mainly characterized by negative associations for NFL and COLIV, which was only observed in typical AD (Figure 3b‐c).ConclusionAtrophy patterns between AD phenotypes showed only minor differences, potentially attributable to the on average later disease stage of the study cohort. The higher immunoreactivity for pathological markers found in atypical AD might suggest a more severe disease burden. (Para)hippocampal volume decline was associated with axonal and microvascular deterioration in typical AD, but not in the higher pathologically burdened atypical AD group, suggesting a differential susceptibility between AD phenotypes.