Karly Alex Cody

The genetic contributions of the X chromosome to Alzheimer’s disease (AD) remain poorly understood yet are expected to importantly shape sex differences in AD. We therefore performed large-scale X-chromosome-wide association studies (N=1,240,451), evaluating differential risk due to sex, APOE *4, and escape from X-chromosome inactivation, finding most X-linked loci appear relevant to female-biased AD etiology. In evaluating genetic pleiotropy with hormonal, lipid, and brain imaging traits, we discovered X-linked AD loci converged on white matter traits, particularly in the anterior corona radiata and splenium of the corpus callosum. Through brain-centric functional genomics analyses, we then nominated candidate causal genes, including 5 that appeared highly robust. Notably, we found the escape gene RBBP7 decreases AD risk in APOE *4 carriers likely through higher expression in excitatory neurons to counter tau-related neurodegeneration. Altogether, we provide an atlas of sex and APOE *4-informed candidate X-linked AD risk loci, genes, and mechanisms that will guide future studies.

Approximately 10% of clinically unimpaired individuals with abnormal amyloid (A+; preclinical Alzheimer’s disease) have ‘divergent’ cortical tau pathology (A+TCortical+), defined as greater than expected tau in cortical regions relative to the medial temporal lobe and/or cortical asymmetry on tau PET in addition to or instead of traditional medial temporal lobe tau burden. Although these A+TCortical+ individuals have subtle cognitive deficits at baseline, the longitudinal imaging and clinical outcomes are unknown. We aimed to characterize longitudinal trajectories of A+TCortical+ individuals compared with other biomarker-defined clinically unimpaired groups, given that identifying those at highest risk for decline is crucial for informing prevention trials and understanding early disease mechanisms. In this longitudinal study, we examined tau PET, MRI, cognitive and functional data from 395 clinically unimpaired participants, aged 65–85 years, enrolled in the Anti-Amyloid Treatment in Asymptomatic Alzheimer’s Disease (A4) Study. Participants had two to five flortaucipir scans over a mean (standard deviation) follow-up period of 4.7 (1.6) years. Changes in regional and voxelwise tau patterns, atrophy, cognition and functioning were examined. Longitudinal trajectories from A+TCortical+ (n = 34) were compared with preclinical Alzheimer’s disease with elevated tau PET signal in the medial temporal lobe only (A+TMTL+, n = 102), preclinical Alzheimer’s disease without significant tau (A+TMTL−, n = 210) and those without amyloid or tau (A−TMTL−, n = 49). Cortical tau accumulation was fastest in A+TCortical+ (0.018–0.034 standardized uptake value ratios per year), whereas medial temporal lobe tau accumulation was comparable across A+TCortical+, A+TMTL+ and A+TMTL− groups (0.010–0.013 standardized uptake value ratios per year). Tau continued to accumulate in affected regions and contralateral homotopic regions in A+TCortical+ participants with asymmetrical tau at baseline such that asymmetrical patterns were maintained over time. Younger A+TCortical+ participants had an especially fast cortical accumulation rate. The A+TCortical+ group showed significantly greater neurodegeneration and faster clinical decline (Clinical Dementia Rating Scale Sum of Boxes = 0.610 points per year; Mini-Mental State Examination = −0.780 points per year) than all other biomarker-defined subgroups (Clinical Dementia Rating Scale Sum of Boxes = 0.048–0.182 points per year; Mini-Mental State Examination = −0.189–0.006 points per year). In summary, individuals with divergent cortical tau patterns continue to accumulate cortical tau at a faster rate, show greater neurodegeneration and have faster cognitive and functional decline than other preclinical Alzheimer’s disease subgroups. Clinical trials and research examining tau progression and clinical decline in preclinical Alzheimer’s disease without subtyping might be disproportionately influenced by this small, high-risk subgroup.

INTRODUCTION: We standardized positron emission tomography (PET) data across multiple cohorts and tracers to characterize the frequency of amyloid and tau PET severity along the clinical continuum. METHODS: Clinical stage was defined using cohort‐specific criteria and included cognitively unimpaired (CU), mild cognitive impairment (MCI), and dementia. Amyloid severity was staged using Centiloids (CL). Tau severity was staged using a hierarchical Braak‐based schema. The cumulative probabilities of PET‐based stages were estimated using ordinal logistic regressions. RESULTS: Among 10,396 individuals (mean [standard deviation] age: 71.9 [7.1] years), amyloid levels ≥ 25 CL increased with age among CU and MCI, while amyloid levels ≥ 100 CL were most common in dementia. In 3295 with tau PET, tau severity increased with amyloid and clinical stage and showed complex associations with age. Within each clinical stage, the full spectrum of amyloid and tau PET severity was observed. DISCUSSION: PET‐based staging revealed heterogeneous amyloid and tau burden along the clinical continuum. Highlights: PET‐based staging is feasible across multiple cohorts and PET tracers. There is heterogeneity in amyloid and tau severity across the clinical spectrum. The frequency of amyloid and tau PET severity increased with clinical severity. The likelihood of tau PET severity differed by age, amyloid, and clinical severity.

Background: This study characterizes the frequency of amyloid and tau PET‐based stages along the clinical continuum in NACC participants from the Alzheimer's Disease Research Center (ADRC) program. Method: 1013 NACC participants with available amyloid and tau PET imaging and diagnosis data were included in this study. Given the clinical heterogeneity present across NACC, parallel analyses were conducted in ADNI for comparison ( n = 926; Table 1). PET data corresponding to ligand‐specific post‐injection windows were processed and harmonized across cohorts using an MRI‐free pipeline. Amyloid PET SUVRs were translated to centiloids (CLs). Amyloid positivity (A+/‐) was defined as 25CL. Regional tau positivity (T+/‐) was derived within ligand using Gaussian mixture modeling. A hierarchical staging schema was used to classify tau positivity (e.g., T‐, T12+, T34+, T56+) and create PET‐based biological stages (e.g., Stage A: A+T‐; Stage B: A+T12+; Stage C: A+T34+; Stage D: A+T56+). The frequency of PET‐based stages was examined across clinical stages and compared between cohorts. Ordinal logistic regression models were used to estimate the cumulative prevalence of biological stages according to age (mean‐centered), cohort, and clinical stage. Result: Compared to ADNI, NACC had twice as many individuals with dementia (Table 1), of which 35% were presumed to be due to non‐AD suspected etiologies (Figure 1A). Across cohorts, tau PET stages generally increased with both amyloid and clinical severity (Figure 1B). Notably, the frequency of biological stages within dementia differed significantly between cohorts such that NACC had a significantly higher proportion of participants in Stage D, while ADNI had more participants in Stages A‐C (χ ² (3)=11.5, p = 0.009; Figure 2A). Analyses estimating the prevalence of biological stages reflected this difference in the distribution of biological stages across clinical stages between cohorts (cohort×clinical stage, p <.001). The estimated prevalence of the biological stages also differed by age and clinical stage (age×clinical stage, p <.001), with younger age associated with higher probability of Stage D in dementia across both cohorts (Figure 2B). Conclusion: Together, these findings underscore heterogeneity in amyloid and tau PET burden along the clinical continuum and highlight the application of amyloid and tau PET‐based staging frameworks in a clinically heterogeneous sample.

INTRODUCTION: Alzheimer’s disease (AD) prevention trials have multiple screening steps to identify cognitively unimpaired individuals with AD biomarker evidence. Cognitive/functional screening tests may be biased in underrepresented groups, thereby impacting trial eligibility. METHODS: 6669 participants screened for the Anti-Amyloid Treatment in Asymptomatic Alzheimer’s (A4) study were grouped by ethnoracial background and testing language. Ethnoracial/language differences in ineligibility reason, cognitive/functional test performance, and amyloid positivity rates were examined. RESULTS: Ethnoracial minorities were least likely to meet eligibility criteria. Patterns of incorrect Mini-Mental State Examination items and impaired Clinical Dementia Rating functional domains differed between groups excluded for impaired cognition/function, suggesting test biases. The Free and Cued Selective Reminding Test yielded more similar exclusion rates across groups than Logical Memory. Cognitive/functional screening biases may impact subsequent biomarker screening as amyloid positivity rates were lowest in ethnoracial minorities. DISCUSSION: Biases in cognitive/functional screening tests may disproportionately exclude ethnoracial minorities in AD prevention trials.

Characterizing the timing and progression of Alzheimer’s disease biomarker onset in Down syndrome (DS) and contrasting potential timing differences with neurotypical adults is needed to identify optimal Alzheimer’s disease therapeutic treatment windows in DS. In this study, 198 adults with DS from the Alzheimer Biomarker Consortium–Down Syndrome and 172 neurotypical adults from the Wisconsin Registry for Alzheimer’s Prevention with available longitudinal beta-amyloid PET, tau PET and plasma p-tau217 analysed on the Lilly Meso Scale Delivery Platform were included. Individuals with DS had a significantly higher lifetime risk of beta-amyloid plaque onset. Temporal modelling of longitudinal biomarker measures revealed earlier age at positivity of beta-amyloid plaques, p-tau217 and neurofibrillary tau tangles in DS relative to the neurotypical cohort. The onset of p-tau217 and tau PET positivity in DS occurred nearly simultaneously, roughly 4–6 years following beta-amyloid onset, whereas the neurotypical group displayed greater temporal latency between positivity of the two biomarkers. The early and simultaneous onset of these biomarkers in DS highlights the necessity for early therapeutic interventions in this population. This work, combined with the upcoming anti-amyloid safety and efficacy clinical trials for DS will help to identify optimal treatment windows for these individuals.

Characterizing the onset and progression of Alzheimer’s disease pathologies relative to one another is important for biological staging and clinical trial design. Recent advances in blood plasma assays of Alzheimer’s disease amyloid and tau pathology have enabled detection of Alzheimer’s disease pathophysiology during life, but it remains unclear when plasma biomarker abnormalities are detectable relative to established amyloid and tau PET imaging biomarkers, and the extent to which plasma biomarkers can be used for biological disease staging. This work applies a novel temporal modelling approach to amyloid PET and plasma p-tau217 data from two different assay platforms to characterize when plasma p-tau217 become abnormal relative to amyloid PET, tau PET, and cognitive decline in a predominantly cognitively unimpaired cohort. This study included a subset of 172 Wisconsin Registry for Alzheimer’s Prevention participants (mean (standard deviation (SD)) age at baseline plasma = 63.2 (6.3) years; 149 cognitively unimpaired at last cognitive assessment) with available amyloid PET imaging and plasma p-tau217 data assayed on the Lilly Meso Scale Delivery and Quanterix Alzpath platforms. We estimated the within-person onsets of detectable amyloid PET and plasma p-tau217 using sampled iterative local approximation and investigated the impact of this timing on downstream tau PET accumulation and cognitive decline using linear mixed-effects models. Longitudinal modelling revealed that on average, amyloid PET positivity preceded p-tau217 positivity, and both amyloid and p-tau217 preceded detectable changes in brain tau accumulation. Comparisons of ‘time from biomarker onset’ indicated that time from p-tau217 onset explained more variability in tau PET accumulation and cognitive decline than time from amyloid PET onset for the Lilly assay but did not differ for the Alzpath assay. Overall, the timing between amyloid PET and p-tau217 onset (in a subset positive for both) ranged from −5.5–24.6 years. These results suggest that plasma p-tau217 follows a predictable path once above thresholds thereby enabling estimation of p-tau217 + age and suggesting these assays may be useful for disease staging. Information regarding the timing of abnormal detection of amyloid PET, plasma p-tau217, and tau PET in relation to preclinical cognitive decline suggests that an optimal window for secondary prevention of Alzheimer’s disease may be within ten years of amyloid PET positivity and within five years of plasma p-tau217 positivity. Future work is needed to identify sources of observed interindividual heterogeneity in the timing of biomarker abnormalities and cognitive decline and impairment following biomarker positivity.

Staging the severity of Alzheimer’s disease pathology using biomarkers is central to early detection and therapeutic trial design. In this cross-sectional study, we standardized amyloid and tau PET data across multiple cohorts to characterize the frequency of amyloid and tau PET-based stages across the clinical continuum. We examined amyloid and tau severity in 10,396 participants (mean [SD] age, 71.9 [7.1] years) with amyloid PET imaging and a subset (n = 3,295) with tau PET imaging. Clinical stage was defined using cohort-specific criteria and categorized as cognitively unimpaired (n = 7,764), mild cognitive impairment (n = 1,480), or dementia (n = 1,152). Amyloid positivity was defined as ≥25 centiloids and amyloid severity was staged using centiloids bins (e.g., <10, 10-24, 25-49, 50-74, 75-99, ≥100). Tau PET severity was staged using a hierarchical Braak staging schema (e.g., T−, T12+, T34+, T56+), and combined with amyloid status to operationalize PET-based Alzheimer’s disease biological stages (e.g., Stage A: A+T−; Stage B: A+T12+; Stage C: A+T34+; Stage D: A+T56+). The cumulative probabilities of PET-based stages were estimated using ordinal logistic regression models. In cognitively unimpaired individuals, the frequency of amyloid levels ≥10 centiloids increased with age. Similarly, amyloid levels ≥25 centiloids increased with age in mild cognitive impairment. Overall, elevated amyloid (≥25 CL) was more likely with increasing age among non-demented individuals. By contrast, this age association was attenuated in dementia where severe amyloid burden (e.g., ≥100 CL) was most common. In the tau PET subsample (n = 3,295), there was a three-way interaction between amyloid, age, and clinical impairment on likelihood of tau severity. Both higher amyloid and greater clinical impairment were associated with increased tau severity; however, the strength and direction of these associations varied with age. At lower amyloid levels, the odds of tau severity increased with older age among cognitively unimpaired and mild cognitive impairment. Conversely, at higher amyloid levels, the odds of higher tau severity (e.g., T56+) decreased with older age in mild cognitive impairment and dementia. A similar age-related pattern was observed in the frequency of biological stages (n = 1,154), where Stage D (e.g., A+T56+) was most frequent in younger individuals with dementia. These findings underscore the dual importance of amyloid and tau PET severity as biomarkers for staging and characterizing Alzheimer’s disease progression. They also demonstrate the feasibility of applying PET-based staging frameworks for the diagnosis of Alzheimer’s disease across multiple tracers and cohorts.

INTRODUCTION: Amyloid positron emission tomography (PET) scans collected by Alzheimer's Disease Research Centers (ADRCs) include site‐specific “mixed protocols” and, since 2021, standardized protocols through the Standardized, Centralized AD/ADRD Neuroimaging initiative. We compared amyloid PET from cohorts using unstandardized versus standardized pre‐processing protocols. METHODS: Across cohorts, amyloid measured in Centiloids (CLs) was derived from 18F‐florbetapir (FBP), 18F‐florbetaben (FBB), 11C‐Pittsburgh compound B (PIB), and 18F‐NAV4694 (NAV) scans using identical magnetic resonance imaging–free post‐processing pipelines. We evaluated CL distributions across cohorts and tracers using Gaussian mixture models (GMMs). RESULTS: Within each tracer, GMM‐derived amyloid negative (A–) CL distributions were similar across cohorts. PIB and NAV showed lower CL variability in the A– range than FBP and FBB. DISCUSSION: Consistent post‐processing yielded comparable A– distributions within tracer across cohorts with heterogeneous pre‐processing protocols, supporting future ADRC PET data integration. Reduced variability in A– classes for PIB and NAV suggest that lower positivity thresholds may be appropriate. Highlights: Alzheimer's Disease Research Center (ADRC) amyloid positron emission tomography (PET) includes unstandardized and standardized pre‐processing. With consistent post‐processing, amyloid‐negative (A–) Centiloids (CLs) were similar despite variable pre‐processing. CLs from 11C‐Pittsburgh compound B and 18F‐NAV4694 were less variable than 18F‐florbetaben and 18F‐florbetapir. Similar A– CLs across variable pre‐processing supports combining ADRC amyloid PET.

The temporal relationship between cerebrovascular disease (V), indicated by white matter hyperintensities, and beta-amyloid (A) in Alzheimer’s disease remains unclear, prompting speculation about their potential interdependence. Longitudinal data were employed to estimate onset ages and corresponding disease chronicity for A and V (where disease chronicity is calculated as age at measurement minus estimated age of biomarker abnormality onset). In a large, predominantly cognitively unimpaired dataset (n = 877, ages 43–93 years), a V+ threshold was identified, and Sampled Iterative Local Approximation (SILA) was utilized to illustrate the predictable accumulation trajectory of V post-onset. Investigating the temporal association between A and V onset ages and accumulation trajectories in preclinical years, four operationalizations of time were examined across two initially cognitively unimpaired samples (n = 240 primary sample from Wisconsin Registry for Alzheimer’s Prevention; n = 123 replication sample from Wisconsin Alzheimer’s Disease Research Center): (i) chronological age, (ii) estimated V+ chronicity, (iii) years since baseline scan, and (iv) estimated A+ chronicity. Results indicated that while both diseases are age-related, their onsets and trajectories are independent of each other. In addition, results indicated that V and A accumulation trajectories were highly predictable relative to onset of positivity for each biomarker. Cognitive decline across multiple cognitive domains was fastest when both V and A were present based on last available amyloid PET and MRI scan, with greater A chronicity being a more salient predictor of cognitive decline in these samples.