Search

Introduction: Hispanic/Latino populations are underrepresented in Alzheimer Disease (AD) genetic studies. Puerto Ricans (PR), a three-way admixed (European, African, and Amerindian) population is the second-largest Hispanic group in the continental US. We aimed to conduct a genome-wide association study (GWAS) and comprehensive analyses to identify novel AD susceptibility loci and characterize known AD genetic risk loci in the PR population. Materials and methods: Our study included Whole Genome Sequencing (WGS) and phenotype data from 648 PR individuals (345 AD, 303 cognitively unimpaired). We used a generalized linear-mixed model adjusting for sex, age, population substructure, and genetic relationship matrix. To infer local ancestry, we merged the dataset with the HGDP/1000G reference panel. Subsequently, we conducted univariate admixture mapping (AM) analysis. Results: We identified suggestive signals within the SLC38A1 and SCN8A genes on chromosome 12q13. This region overlaps with an area of linkage of AD in previous studies (12q13) in independent data sets further supporting. Univariate African AM analysis identified one suggestive ancestral block (p = 7.2x10-6) located in the same region. The ancestry-aware approach showed that this region has both European and African ancestral backgrounds and both contributing to the risk in this region. We also replicated 11 different known AD loci -including APOE- identified in mostly European studies, which is likely due to the high European background of the PR population. Conclusion: PR GWAS and AM analysis identified a suggestive AD risk locus on chromosome 12, which includes the SLC38A1 and SCN8A genes. Our findings demonstrate the importance of designing GWAS and ancestry-aware approaches and including underrepresented populations in genetic studies of AD. [ABSTRACT FROM AUTHOR]

INTRODUCTION: Neuropsychiatric symptoms (NPS) are highly prevalent in Alzheimer's disease (AD). There are no effective treatments targeting these symptoms. METHODS: To facilitate identification of causative mechanistic pathways, we initiated an effort (NIH: U01AG079850) to collate, harmonize, and analyze all available NPS data (≈ 100,000 samples) of diverse ancestries with whole‐genome sequencing data from the Alzheimer's Disease Sequencing Project (ADSP). RESULTS: This study will generate a genomic resource for Alzheimer's disease with both harmonized whole‐genome sequencing and NPS phenotype data that will be publicly available through NIAGADS. Primary analyses will (1) identify novel genetic risk factors associated with NPS in AD, (2) characterize the shared genetic architecture of NPS in AD and primary psychiatric disorders, and (3) assess the role of ancestry effects in the etiology of NPS in AD. DISCUSSION: Expansion of the ADSP to harmonize and refine NPS phenotypes coupled with the proposed core analyses will lay the foundation to disentangle the molecular mechanisms underlying these detrimental symptoms in AD in diverse populations. Highlights: Neuropsychiatric symptoms (NPS) are highly prevalent in Alzheimer's disease (AD).There are no effective treatments targeting NPS in AD.The current effort aims to collate, harmonize, and analyze all NPS data from the Alzheimer's Disease Sequencing Project.Core analyses will identify underlying genetic factors and mechanistic pathways.The harmonized genomic and phenotypic data from this initiative will be available through National Institute on Aging Genetics of Alzheimer's Disease Data Storage Site. [ABSTRACT FROM AUTHOR]

Background: The "Recruitment and Retention for Alzheimer's Disease Diversity Genetic Cohorts in the ADSP (READD‐ADSP)" is developing a resource to expand ancestral diversity in Alzheimer disease (AD) studies to dissect the genetic architecture of AD across different populations. In addition to US sites, READD‐ADSP includes four US sites and nine countries in sub‐Saharan Africa through the Africa Dementia Consortium (AfDC). The overall goal of READD‐ADSP is to identify genetically driven targets in diverse groups including African Americans and Hispanic/Latinos in US, and Africans. In this preliminary analysis we investigated the ancestral genetic differences and the impact of known AD risk factors within West African cohorts. Method: Genome‐wide genotyping was performed on 91 AD cases and 97 cognitive unimpaired controls from Nigeria and Ghana. APOE alleles and ABCA7 deletion (rs142076058) were sequenced using Sanger. We calculated global ancestry (principal components) using the PC‐AiR approach that is robust to known and cryptic relatedness. We investigated known AD loci from non‐Hispanic White (NHW) and AA genome wide association studies. For association analysis, we employed a mixed‐model regression approach (SAIGE) where we controlled for age, gender, population substructure (first three principal components), and relatedness. Result: Principal component analysis identified a distinction between the Ghana and Nigerian cohorts along the first principal component (PC1). Among the genetic loci examined, several showed nominal significance. Notably, the most prominent marker was found in SORL1 (rs17125523; p = 2 × 10‐3). Additionally, we discovered an exonic nonsynonymous marker in the BIN1 gene (rs112318500), which is specific to African ancestry and showed a protective effect. APOE e4 allele showed a significant association with AD risk (OR = 2.5; CI:1.5‐4.2; pv = 0.001), while the e2 indicated a protective trend but did not reach statistical significance. No statistical difference in the frequency of ABCA7 deletion was observed between AD and CU individuals. Conclusion: Our findings highlight the presence of genetic variations between West African populations that warrant further investigation, potentially offering new insights into the genetic underpinnings of AD. Data collection is ongoing across the AfDC and updated data will be presented. [ABSTRACT FROM AUTHOR]

Background: Despite its high heritability, the genetic mechanisms influencing Alzheimer's Disease (AD), particularly in health disparity populations like African Americans (AA) and Hispanics (HI), are not fully understood. The lack of ancestral diversity in genetic datasets, notably in eQTL studies that associate genetic variation with gene expression, exacerbates these disparities. Our study seeks to address this gap by comparing the AD interactions of racially and ethnically diverse expression Quantitative Trait Loci (eQTL) effects to investigate the genetic influence on AD in underrepresented populations. Method: We investigated the impact of AD status on multi‐ethnic eQTLs across 3 diverse ancestral cohorts: AA, HI, and Non‐Hispanic White (NHW). Genotype and gene expression data were collected from samples of each genetic ancestry (AA = 224, HI = 293, NHW = 235), all with known AD status, with a focus on generating whole blood RNAseq data. We modified the TensorQTL pipeline to incorporate an interaction term for AD status to identify eQTL effects potentially induced by disease. We examined the top eQTL association per gene to assess the impact of AD status across the three ancestral cohorts. Result: Our preliminary analysis reveals striking variations in the impact of AD status on gene expression across AA, HI, and NHW cohorts. Several genes in each ancestry show statistically significant (pv<10e‐8) interaction effects (G X AD) (AA = 11, HI = 5, NHW = 9), indicating that genetic variation within those genes influences gene expression in an AD‐status dependent manner. A number of genes displayed interesting interaction patterns, where the most significant eQTL‐AD interaction was identical in two of the three ancestries studied, suggesting a potential ancestrally‐shared connection to AD risk (AA+HI = 4, AA+NHW = 3, HI+NHW = 5, AA+HI+NHW = 0), though notably there were no genes with induced eQTLs across all three ancestries. Conclusion: Our study highlights the impact the AD disease process has on genetic variants that alter whole‐blood gene expression. We also highlight differences by population, prompting the development of ethnically diverse gene expression panels that pave the way for more accurate genetic studies in health disparity populations. Our findings also underscore the potential for discovering novel genetic mechanisms influencing AD risk, which could lead to more effective and inclusive therapeutic strategies. [ABSTRACT FROM AUTHOR]

Background: The utility of neuropsychological measures commonly used to diagnose individuals with suspected MCI or AD was recently explored in European cohorts, however their utility in Caribbean Hispanic (CH) populations is not well understood. Method: Our sample consisted of 507 CH individuals from ongoing studies of Puerto Rican and Cuban American older adults (74%% Female, mean age = 74.1, mean education = 12.7 years, 209 Unaffected, 185 MCI, 113 with AD) who completed cognitive measures from the NACC UDS2 battery (Logical Memory, Semantic Fluency [animals and vegetables], and Digit Span). Participants were clinically adjudicated as unaffected, MCI, or AD, and had a CDR global score ≤1. Individuals with disorders known to affect cognition were excluded. An ordinal regression explored which cognitive tests were most strongly associated with clinical diagnoses. We ran this analysis again on a subset of participants who also completed the Trails Making Test (n = 242, 103 unaffected, 97 MCI, 42 AD). We examined both the log‐odds estimate and Wald value. Result: An ordinal regression on the full sample showed that the tests associated with differences in clinical diagnosis in order of their predictive value are Delayed Logical Memory (log‐odds = 1.29, Wald = 40.1, p<0.001), Digit Span Backwards (log‐odds = 0.70, Wald = 25.2, p<0.001), Animals (log‐odds = 0.44, Wald = 12.0, p<0.001), Vegetables (log‐odds = 0.32, Wald = 7.3, p<0.01), and Digit Span Forwards (log‐odds = 0.24, Wald = 6.3, p = 0.012). On the subset that completed the Trails Making Test, Trails B emerged as most predictive (log‐odds = 0.70, Wald = 26.8, p <0.001), but Vegetables (p = 0.26), Logical Memory Immediate (p = 0.43), and Trails A (p = 0.70) were not significant predictors of differences in clinical diagnosis in this smaller group. Conclusion: In our sample of CH individuals, we found performance on measures of delayed contextual memory (Logical Memory), working memory (Digit Span Backwards and Trails B) and semantic fluency (Animals) to be most useful in differentiating individuals clinically diagnosed as unaffected, MCI, or AD. These results highlight possible performance patterns specific to CH that can inform adjudication efforts with this population. [ABSTRACT FROM AUTHOR]

Background: Prior studies have shown differences in the genetic etiology and clinical presentation of Alzheimer's Disease across populations. For example, for multiple genetic loci associated with AD, effect sizes can vary drastically between individuals of different ancestral backgrounds. Few investigations into differences in epigenetic features like DNA methylation have been conducted in AD, particularly in diverse individuals. These studies are critical to identify and further characterize mechanisms of disease allowing for development of therapeutic interventions. Method: As part of an ongoing study of the genetics and epigenetics of AD in diverse populations, we performed a methylome analysis of 626 individuals. DNA from whole blood was analyzed using the Illumina MethylationEPICv2.0. The cohort consisted of both AD and cognitively unimpaired (CU) individuals of European (68 AD, 67 CU), African (98 AD, 106 CU), or Hispanic (Puerto Rican (85 AD, 76 CU); Peruvian (41 AD, 41 CU); Cuban (22 AD, 22 CU)) backgrounds. We analyzed data using the SeSAMe R package for quality control and statistical analysis. We performed differential methylation analysis between AD and CU in the overall dataset and within each ancestral population using linear models with covariates sex, age of exam, batch effect, global ancestry and estimated immune cell type proportions. Result: 878,853 CpG sites were tested for differences between AD status. In a preliminary analysis of these data, we identified 563 CpG sites with nominally significant differences (p‐value ≤ 0.001) between AD and CU. Within each ancestral group, the number of differentially methylated sites differed: European – 442 sites, African – 217 sites, Hispanic – 475. Notably, however, the markers within the ancestral group did not overlap, implying that the AD disease process may be quite different across populations. Conclusion: While these results are preliminary, and expansion of the dataset may reveal convergence of methylation patterns across ancestral populations, our analyses suggest the possibility of ancestry specific whole blood DNA methylation patterns as signatures of AD pathogenesis. Ultimately, combining these methylation profiles with existing genomic and transcriptomic data may reveal distinct genes but similar underlying pathological processes contributing to AD across individuals of diverse ancestries. [ABSTRACT FROM AUTHOR]

Background: The ADSP is a National Institute on Aging (NIA) initiative focused on identifying genetic risk and protective variants for Alzheimer Disease (AD). Initial phases (Discovery and Discovery Extension) were predominantly non‐Hispanic Whites of European Ancestry (NHW‐EA). The ADSP expanded the population diversity in the Follow Up Study (ADSP‐FUS), and the current phase, ADSP‐FUS 2.0: The Diverse Population Initiative, focusing on whole genome sequencing (WGS) of non‐European populations including Hispanic/Latino (HL), non‐Hispanic Black with African Ancestry (NHB‐AA) and Asian populations. Support for these efforts include newly funded initiatives such as The DAWN Project, focused on recruitment of African, African‐American and Hispanic American populations, and the Asian Cohort for Alzheimer's Disease (ACAD). Methods: ADSP cohorts consist of studies of AD, dementia, and age‐related conditions. Clinical classifications are assigned based on standard criteria and derived from clinical measures and history, as well as additional neuropathologic data. In addition to production of WGS, APOE genotyping is available for all ADSP samples. Results: The ADSP currently consists of 40 cohorts comprised of ∼36,300 individuals, with plans to sequence >110,000 individuals from diverse race/ethnicity. Genotyping, sequencing, and clinical adjudication has been performed on 36,361 participants (cases N = 12,133, median age = 72; cognitively‐unimpaired(CU) individuals N = 17,116, median age = 74; ADRD N = 7,112, median age = 71). Mean ages for cases and controls vary across cohorts, 57.0+5.6 to 86.5+4.2 cases and 63.3+7.8 to 90.0+0 controls. 61% participants female, distributed as follows: cases(60.3%), CU(63.7%), and ADRD(55.8%). APOE genotype proportions differ considerably across reported race/ethnicity, for example highest for APOE ε4/ε4 carriers observed in Non‐Hispanic whites participants (7.4%) and the lowest in Asians (1.7%) Conclusion: The results provide an overview of clinical features in ADSP cohorts. The growth of the ADSP‐FUS 2.0 is central to the ADSP and expanding the size and diversity of this genomic resource available via NIAGADS. WGS data will be integrated with ADSP programs focused on phenotype harmonization, association analyses, functional genomics, and machine learning. In concert with these programs, the ADSP‐FUS 2.0 will accelerate the identification and understanding of potential genetic risk and protective variants for AD across all populations with the target of developing new treatments that are globally effective. [ABSTRACT FROM AUTHOR]

Background: Polygenic Risk Scores (PRS) are important in predicting disease risk and are usually rely on markers selected by thresholding p‐values from genome‐wide association studies (GWAS). In traditional approaches, one single model is built to calculate risk scores, employing effect size to determine additive risk. However, this traditional method overlooks potential interactions between genetic loci resulting in reduced prediction power. To overcome these limitations, we propose an interpretable machine learning approach based on pathway‐level PRS (p‐PRS). Our approach improves prediction accuracy of PRS and generates both pathway specific and overall PRS considering possible nonlinear interactions. This advancement opens new avenues in personalized medicine, enabling more accurate disease prediction and prevention strategies. Method: We used whole genome sequencing data from 652 individuals from the Puerto Rican Alzheimer's Disease Initiative (PRADI). First, we selected Alzheimer disease (AD)‐related pathways and genes based on those reported in the largest non‐Hispanic White AD GWAS to date (Bellinguez et al. 2021). Then, we applied the Clumping and Thresholding (C+T) PRS approach using Bellinguez et al. summary statistics within each pathway (±20Kb for each gene). Finally, we used random forest with pathway‐level PRS values as input to classify AD vs. control. We assessed the performance of the p‐PRS model based on area under the receiver operating characteristic curve (AUC) using a 70%/30% split for training/testing. The outcomes were compared to a traditional C+T PRS model (AD∼PRS+APOE). Result: p‐PRS improves the performance of the traditional PRS model by 4.2% (p = 5.54E‐32). Performance is centered around AUC = 0.689(SE = 0.0022) for p‐PRS and 0.647(SE = 0.0024) for traditional PRS. PRS by pathway enables determining individual‐level important pathways for accurate classification. Top pathways obtained are related to processes implicated in cognitive deficit, neuroinflammation, neurodegeneration and vesicle mediation. Conclusion: Utilizing the innovative p‐PRS approach has improved the estimation of AD risk in the PR cohort. The application of interpretable machine learning approaches allows identifying the most relevant pathways for effective risk prediction and classification. Importantly, improved precision will provide more effective actionable risk mitigation strategies, optimize the selection process for clinical trials, and contribute to the development of more personalized treatment interventions. [ABSTRACT FROM AUTHOR]

Background: Hispanic/Latino populations are underrepresented in Alzheimer Disease (AD) genetic studies. The Puerto Rican (PR) population, a three‐way admixed (European, African, and Amerindian) population is the second‐largest Hispanic group in the continental US. We performed a genome‐wide association study (GWAS) in the PR population to identify novel AD susceptibility loci and characterize known AD genetic risk loci. Method: 652 individuals (349 AD; 303 cognitively unimpaired), ascertained through the Puerto Rico Alzheimer Disease Initiative (PRADI), were included in the analyses. We performed GWAS on the Whole Genome Sequencing (WGS) dataset with a generalized linear‐mixed model adjusting for sex, age, and population substructure as fixed effects and genetic relationship matrix as a random effect. To infer local ancestry, we merged the target PR dataset with appropriate population samples from the HGDP and 1000G reference panels. Subsequently, we conducted univariate admixture mapping (AM) analysis. We also assessed the polygenic risk score (PRS) using the effect sizes from summary statistics from the non‐Hispanic White (NHW) study. Result: We identified a suggestive significant (p<5 × 10‐6) signal (rs11183403; P = 4 × 10‐6) within the SLC38A1 gene on chromosome 12. Univariate African AM analysis identified one suggestive (p<4 × 10‐5) ancestral block located in the same region on chromosomes 12q13.1 (p = 7.2 × 10‐6). We replicated eight known AD loci– APOE, ABCA7, CLU, FERMT2, GRN, PRDM7, SEC61G, and TREM2. Admixture analysis revealed proportions of 68% European, 20% African, and 12% Amerindian in the PR cohort. NHW‐derived PRS has a good prediction power (AUC = 0.62) in the PR dataset. Conclusion: PR GWAS and AM identified a suggestive AD risk locus in the SLC38A1 gene. This region overlaps with an area of linkage of AD in previous studies (12q13). The SLC38A1 gene is associated with ischemic brain damage and its transcription is affected by amyloid‐beta peptide. Our study replicated 8 known AD loci previously identified in European studies and showed good predictive power with NHW‐derived PRS which is likely due to the high European background of the PR population. Including underrepresented populations in genetic studies is important for identifying health disparities and developing effective treatments for AD in all groups. [ABSTRACT FROM AUTHOR]

Background: Over 13,000 individuals from both domestic and African sites will be collected for the READD‐ADSP study. Adjudicating this number of individuals is challenging, so we evaluated knowledge‐based decision tree algorithms to predict clinical diagnoses using nationally representative norms and standard cut‐offs. Additional models were constructed using culturally adjusted cut‐offs, domain average cut‐offs, and exclusion of the Trail Making Test (TMT) which performed poorly. Our aim was to evaluate the accuracy of these decision tree models in the READD‐ADSP dataset. Method: Clinically adjudicated diagnoses by at least two neuropsychologists and/or neurologists were available for 150 individuals from the READD‐ADSP sample (44%‐Unaffected, 36%‐MCI, 9.3%‐AD, 2.7%‐Dementia not AD, 8%‐Cognitively Impaired– Vascular; 19% Hispanic, 81% African‐American, 76%‐Female, Age: 73.0 years, Education: 13.7 years). Eight models were used to classify our sample into the diagnostic categories above and varied in test cut‐offs (standard vs. culturally adjusted), use of domain average cut‐offs (no average vs. average), and use of TMT (inclusion vs. exclusion). Model accuracy was assessed using the F1 score (which balances precision and sensitivity). F1 scores >0.9 are considered excellent, >0.8 are good, and those between 0.6‐0.8 are acceptable. Result: The two models that used culturally adjusted cut‐offs and domain average cut‐offs performed best (with TMT, F1 = 0.81; without TMT, F1 = 0.78). All other models had F1 scores ranging from 0.63‐0.73. While the two best models were similarly accurate based on F1 scores, the difference in the frequency of diagnoses between them was notable. In the model that excluded TMT, 38% of the sample were classified as Unaffected vs. 30% in the model that included TMT; (41% and 47% were classified MCI, respectively). Conclusion: Our findings show that culturally sensitive adjusted cut‐offs and inclusion of domain average cut‐offs improved the accuracy of decision tree model classifications. These findings suggest that implementation of decision tree model approaches in the READD‐ADSP dataset can accurately and efficiently classify individuals. Finally, we plan improve overall accuracy by using a hybrid method where models classify individuals who are clearly Unaffected or AD while more complex cases will be assigned to human adjudication. [ABSTRACT FROM AUTHOR]

Background: Common and rare variants in SORL1 have been associated with increased risk of Alzheimer's disease (AD). Since 2019, we have run an international collaborative research initiative to ascertain a Peruvian cohort for Alzheimer's disease and other related dementias for genetic studies (PeADI). Method: A Peruvian family (4 AD cases and two mild cognitive impairment (MCI) cases) was recruited through the PeADI study. All six family‐member completed a full cognitive assessment, had plasma‐based biomarkers pTau181 and Aβ42/40 measured via SIMOA chemistry on the Quanterix HD‐X, and underwent whole genome sequencing. Variants within AD risk genes as determined by the ADSP Gene Verification Committee were prioritized and variant interpretation was performed according to ACMG recommendations. Result: We identified a SORL1 c.5019G>A (p.Trp1673Ter) variant of Amerindian background in the four AD diagnosed siblings within this family. The two MCI cases did not carry the novel variant. The identified SORL1 variant corresponded to a heterozygous stop‐gain variant in exon 36 replacing tryptophan by a stop codon at position 1673 of the SORL1 protein. In‐silico analysis predicts this variant promotes nonsense‐mediated mRNA decay. This variant has not been previously reported in databases including gnomAD, LOVD and ClinVar. The 4 AD cases had on average 2.3X higher plasma pTau181 concentrations compared to the 2 MCI (2.03 ± 0.28pg/µl vs 0.88 ± 0.7pg/µl). There was no noticeable difference in the Aβ42/40 ratio. This variant is classified as likely pathogenic according to ACMG. Conclusion: We report the first Peruvian AD family carrying a likely pathogenic stop‐gain SORL1 variant within an Amerindian background region. Further cosegregation and functional assays are required to establish the risk size of this variant for AD. [ABSTRACT FROM AUTHOR]

Background: In the US, African Americans (admixed with African and European) followed by Hispanics (admixed with Amerindian, African, and European) are the most affected groups compared to non‐Hispanic Whites (NHW). While genetic diversity and admixture play crucial roles in disease risk, the ancestry‐specific mechanisms remain poorly understood with most AD‐related studies focusing on NHW. Despite the recent field efforts to include genetically admixed populations, there continues to be a lack of functional studies in AD across the different cell types in these populations. Given the importance of Microglia in AD, we here characterize the genetic regulatory architecture (GRA) on iPSC‐derived Microglia (MGL) in African and Amerindian genomes. Method: iPSC lines derived from controls and AD patients with >90% genomic content from different ancestries (Amerindian, African, and European) were differentiated into MGL. We performed bulk RNA‐seq and ATAC‐seq, followed by differential expression and accessibility analyses to study the GRA of these admixed populations and its contributions to AD. Result: We identified 1,103 differentially expressed genes (DEGs) and 267 differentially accessible genes (DAGs) across ancestries. We observed the most differences on both chromatin accessibility and gene expression levels between AI and AF. On the chromatin level and in the context of AD, we observed 2 DAGs (PRDM7 and SCIMP) between AI and AF, and 1 DAG between AI and EU (PRDM7). In addition, we identified 10 AD‐risk modifying genes that are differentially expressed between AI and AF ancestries (ABI3, CTSB, JAZF1, MS4A6A, PILRA, PLEKHA1, RASGEF1C, SORL1, TREM2, and TREML2) and 3 DEGs between AI and EU (JAZF1, MS4A6A, and SORL1). We identified several DEGs to be involved in lipid metabolism, cholesterol biosynthesis and metabolism, lysosomal activity, and immune response ‐ all highly relevant processes in AD pathology. Conclusion: We provide new insights into ancestry‐specific genetic risk factors in AD pathophysiology. Here, we report novel transcriptomic and chromatin accessibility data in microglia of AI and AF ancestries that potentially contribute to a differential genetic risk in AD in the different ancestries. Interestingly, those ancestries with greatest migratory differences revealed the largest DEG. [ABSTRACT FROM AUTHOR]

Background: The Genome Center for Alzheimer's Disease (GCAD) coordinates the integration and meta‐analysis of all available Alzheimer's disease (AD) relevant whole genome sequencing (WGS) data to facilitate the goal of identifying AD risk or protective genetic variants and eventual therapeutic targets. The WGS datasets are generated via the collaboration of scientists from the Alzheimer's Disease Sequencing Project (ADSP) and GCAD. To minimize data heterogeneity introduced by different sequencing protocols and machines, GCAD processes all samples using identical pipelines. Methods: The raw sequencing data are first mapped to GRCh38/hg38 and variants (SNVs and indels) are called using GATK. Additionally, compact VCF and GDS formatted files are generated to facilitate researchers who want to use smaller pVCFs. SNVs and indels are annotated using the ADSP annotation pipeline. Lastly, structural variants (SV) are called using Smoove and Manta and joint genotyped using GraphTyper2. Results: The dataset (ADSP Release 5, R5, 2024) includes ∼60,000 genomes from >50 diverse cohorts with 4 major ancestries: 47% Non‐Hispanic White, 29% Hispanic or Latino, 16% Black or African American and 8% Asian. Data are deeply sequenced (average genome coverage: >30x). CRAMs, gVCFs from GATK, and SV VCFs of a subset of the R5 samples (n = 36,361) were deposited into NIAGADS Data Sharing Service (DSS) (https://dss.niagads.org/) for public distribution in 2022, and similarly, the new samples in R5 will be released after the joint call is complete. In addition, joint‐genotype VCFs on SNVs, indels, and SVs will be available. These will undergo full quality control and annotation process. Conclusion: The ADSP and GCAD generate high quality genotype and SV calls. Currently the project is processing ∼60,000 WGS samples sequenced primarily through the ADSP Follow‐Up Study, which will contain a more ancestrally diverse set of populations. We anticipate this 2024 release will continue to benefit the research community studying AD genetics. [ABSTRACT FROM AUTHOR]

Background: Increasing ethnic/ancestral diversity in genetic studies is critical for defining the genetic architecture of Alzheimer disease (AD). Amerindian (AI) populations are substantially underrepresented in AD genetic studies. The Peruvian (PE) population, with up to ∼80% of AI ancestry, provides a unique opportunity to assess the role of AI ancestry in AD. We performed the first genome‐wide association study (GWAS) in the PE population to identify novel AD susceptibility loci and characterize known AD genetic risk loci. Method: The PE dataset includes array‐genotype and phenotype data from 542 individuals (189 cases; 353 controls), imputed to the NHLBI TOPMedv5 haplotype reference panel. We used a generalized linear mixed‐model (SAIGE software) for the GWAS analysis. We analyzed two separate models; the first model accounted for sex, age, and population substructure, while the second model also included the dosage of APOEe4. In both models, we included a genetic relationship matrix as a random effect to account for any potential relatedness. To determine if the associations are specific to specific ancestries, we employed ancestry‐aware approaches using the RFMix software. Result: APOE was significantly associated with AD with an effect size comparable to that found in non‐Hispanic white (NHW) populations (OR = 3.3(2.2‐4.8),pv = 8.0×10−10). Two additional known AD loci, TREML2 (pv = 0.008) and CLU (pv = 0.012), showed nominal significance Variants at three additional loci reached suggestive significance (pv<1×10−6): NFASC (pv = 9.4×10−8;chromosome 1), STK32A (pv = 9.3×10−7; chromosome 5), and LOC100132830 (pv = 6.7×10−7;chromosome 6). The NFASC locus neared genome‐wide significance in the APOE adjusted model (pv = 6.7×10−8). The haplotypes associated with AD at the NFASC locus were found to be of European origin. Additionally, the STK32A locus was found to have a protective effect specifically among individuals of AI background. We did not observe significant heterogeneity of effect at the APOE and LOC100132830 loci across different ancestral backgrounds. Conclusion: PE GWAS identified a novel, promising AD susceptibility locus in the NFASC gene of European origin. We also detected a potential protective effect in the STK32A locus on AI background, emphasizing the importance of incorporating ancestry‐aware approaches in gene discovery in admixed populations. [ABSTRACT FROM AUTHOR]

Background: Neuropsychiatric Symptoms (NPS) including aggression, psychosis, anxiety, apathy and depression are highly prevalent in Alzheimer's Disease patients and are associated with accelerated decline and a detrimental impact on suffering and quality of life of both patients and caregivers. There are no effective pharmaceutical interventions targeting these symptoms, making a better understanding of the etiologic mechanisms underlying NPS in AD critical to develop improved treatments. Method: To facilitate identification of genetic loci and mechanistic pathways underlying NPS in AD, we have initiated an effort (NIH: U01AG079850) to collate and harmonize all available NPS data in over 70 cohorts (>80,000 samples) of diverse ancestries with whole‐genome sequencing data from the Alzheimer's Disease Sequencing Project (ADSP), and analyze these data to identify genetic loci and mechanistic pathways associated with NPS in AD. Result: A publicly available genomics resource for NPS in AD with extensive harmonized NPS phenotype data. Primary core analyses will (1) identify novel genetic risk factors associated with neuropsychiatric symptoms in AD, (2) characterize the shared genetic architecture of neuropsychiatric symptoms in AD and primary psychiatric disorders, and (3) assess the role of ancestry effects in the etiology of neuropsychiatric symptoms in AD. Conclusion: Expansion of the ADSP to harmonized and refined NPS phenotypes coupled with the proposed core analyses will lay the foundation to disentangle the molecular mechanisms underlying these detrimental symptoms in AD in diverse populations. [ABSTRACT FROM AUTHOR]

Background: Alzheimer disease (AD) is the fourth leading cause of death in Puerto Rico. The Puerto Rican (PR) population has a high proportion of older adults (18%, over > 65), with 12.5% of them suffering from ADThese statistics highlight the need to investigate the genetic risk factors underlying AD in the PR population, as it could lead to the development of targeted treatments and therapies. Moreover, the ancestrally admixed makeup of the PR population provides an opportunity to assess the role of the European (∼67%), African (∼20%) and Amerindian (∼13%) ancestry in AD risk. We performed genome wide association analysis (GWAS) using whole genome sequence data to identify genetic risk/protective factors associated with AD in the PR population. Method: The PR dataset includes WGS and phenotype data from 640 individuals, comprising 335 AD and 305 cognitively unimpaired (CU) controls. To account for the population substructure, we calculated the global ancestry (principal components) using EIGENSTRAT. We performed GWAS analyses with a generalized linear mixed‐model using the SAIGE software. The model included genotype, sex, age, and principal components (population substructure) as fixed effects and genetic relationship matrix as a random effect. The genetic relationship matrix was calculated based on genomic data and accounted for the relatedness among the individuals in the dataset. Result: We identified four suggestive significant loci (P<1×10−6) associated with the risk of AD in PRs: NACC2 (pv = 4.8×10−7) on chromosome 9, SCN8A (pv = 9.3×10−7) on chromosome 12, FOXK2 (pv = 9.9×10−7) on chromosome 17, and APOEe4 (pv = 6.8×10−8) on chromosome 19. Eight additional AD loci with the same lead marker from European GWAS study (Bellenquez et al.) showed nominal significance: FERMT2 (pv = 6.2×10−3), TREM2 (pv = 8.0×10−3), CLU (pv = 1.8×10−2), RASGEF1C (pv = 2.4×10−2), ADAM17 (pv = 3.5×10−2), DOC2A (pv = 4.3×10−2), GRN (pv = 4.4×10−2) and SORL1 (pv = 5.1×10−2). Conclusion: This study identified three suggestive novel significant loci (NACC2, SCN8A, and FOXK2) associated with AD risk in PRs. In addition, GWAS study on PRs with a high proportion of European ancestry was able to replicate nine AD loci previously identified in European studies. These findings provide new insights into the genetic architecture of AD in the PR population. [ABSTRACT FROM AUTHOR]

Background: The Genome Center for Alzheimer's Disease (GCAD) coordinates the integration of all available Alzheimer's disease (AD) relevant whole genome sequencing (WGS) data with the goal of identifying AD risk or protective genetic variants and eventual therapeutic targets. The WGS datasets are generated through collaboration between investigators from the Alzheimer's Disease Sequencing Project (ADSP) and GCAD. With the goal of minimizing data heterogeneity, introduced by different sequencing protocols and assays, GCAD processes all samples using standardized pipelines and performs quality control (QC)/quality assurance (QA) checks. Methods: Raw sequencing data (FASTQs or BAMs) were aligned to GRCh38/hg38 by BWA, and variant calling and joint genotyping on single nucleotide variants (SNVs), insertions and deletions (indels), were done by GATK. Structural variants (SVs) were called per sample using the Smoove, Manta, and Strelka packages. Preliminary QA checks including sex check, contamination, and genotype concordance were performed followed by QC per ADSP protocol to evaluate the quality of samples and variants. To facilitate access and usage of massive joint‐genotype called VCF files, a compact version for storing variant info and sample genotypes only was released first. Results: We dropped 275 (0.7%) samples of poor coverage (<20×), and we flagged 219 (0.6%) samples that were of borderline quality. As a result, the dataset (ADSP Release 4, 2022) includes 36,361 genomes from 40 diverse cohorts with 4 major ancestries: 16,573 Non‐Hispanic Whites, 11,358 Hispanics; 5,422 African Americans; and 2,802 Asians. Data are deeply sequenced (average genome coverage: 40x). All samples' CRAMs and gVCFs from GATK were deposited into NIAGADS Data Sharing Service (DSS) (https://dss.niagads.org/) for public distribution. Joint‐genotyped called VCFs are undergoing a full QC/annotation process and will be made available. This joint‐genotyped called VCF contains >362M bi‐allelic variants, >58M multi‐allelic variants, with 95% of variants remaining after QC. SV calling is ongoing and data will be ready prior to the conference. Conclusion: The ADSP and GCAD generate high quality SNVs, indels and SV calls. Currently GCAD is preparing the next release of ∼60,000 more ancestrally‐diverse WGS samples sequenced primarily through the ADSP Follow‐Up Study, which we anticipate will be released in 2023 to greatly benefit the AD genetics community. [ABSTRACT FROM AUTHOR]

Introduction: Genome-wide association studies (GWAS) studies in Alzheimer's disease (AD) demonstrate ancestry-specific loci. Previous studies in the regulatory architecture have only been conducted in Europeans (EUs), thus studies in additional ancestries are needed. Given the prevalence of AD genes expressed in microglia, we initiated our studies in induced pluripotent stem cell (iPSC) -derived microglia., Methods: We created iPSC-derived microglia from 13 individuals of either high Amerindian (AI), African (AF), or EU global ancestry, including both AD and controls. RNA-seq, ATAC-seq, and pathway analyses were compared between ancestries in both AD and non-AD genes., Results: Twelve AD genes were differentially expressed genes (DEGs) and/or accessible between ancestries, including ABI3, CTSB, and MS4A6A. A total of 5% of all genes had differential ancestral expression, but differences in accessibility were less than 1%. The DEGs were enriched in known AD pathways., Discussion: This resource will be valuable in evaluating AD in admixed populations and other neurological disorders and understanding the AD risk differences between populations., Highlights: First comparison of the genomics of AI, AF, and EU microglia. Report differences in expression and accessibility of AD genes between ancestries. Ancestral expression differences are greater than differences in accessibility. Good transcriptome correlation was seen between brain and iPSC-derived microglia. Differentially expressed AD genes were in known AD pathways., (© 2025 The Author(s). Alzheimer's & Dementia published by Wiley Periodicals LLC on behalf of Alzheimer's Association.)

The Alzheimer's Disease Sequencing Project (ADSP) is a national initiative to understand the genetic architecture of Alzheimer's Disease and Related Dementias (AD/ADRD) by sequencing whole genomes of affected participants and age-matched cognitive controls from diverse populations. The Genome Center for Alzheimer's Disease (GCAD) processed whole-genome sequencing data from 36,361 ADSP participants, including 35,014 genetically unique participants of which 45% are from non-European ancestry, across 17 cohorts in 14 countries in this fourth release (R4). This sequencing effort identified 387 million bi-allelic variants, 42 million short insertions/deletions, and 2.2 million structural variants. Annotations and quality control data are available for all variants and samples. Additionally, detailed phenotypes from 15,927 participants across 10 domains are also provided. A linkage disequilibrium panel was created using unrelated AD cases and controls. Researchers can access and analyze the genetic data via NIAGADS Data Sharing Service, the VariXam tool, or NIAGADS GenomicsDB.

Introduction: Neuropsychiatric symptoms (NPS) are highly prevalent in Alzheimer's disease (AD). There are no effective treatments targeting these symptoms., Methods: To facilitate identification of causative mechanistic pathways, we initiated an effort (NIH: U01AG079850) to collate, harmonize, and analyze all available NPS data (≈ 100,000 samples) of diverse ancestries with whole-genome sequencing data from the Alzheimer's Disease Sequencing Project (ADSP)., Results: This study will generate a genomic resource for Alzheimer's disease with both harmonized whole-genome sequencing and NPS phenotype data that will be publicly available through NIAGADS. Primary analyses will (1) identify novel genetic risk factors associated with NPS in AD, (2) characterize the shared genetic architecture of NPS in AD and primary psychiatric disorders, and (3) assess the role of ancestry effects in the etiology of NPS in AD., Discussion: Expansion of the ADSP to harmonize and refine NPS phenotypes coupled with the proposed core analyses will lay the foundation to disentangle the molecular mechanisms underlying these detrimental symptoms in AD in diverse populations., Highlights: Neuropsychiatric symptoms (NPS) are highly prevalent in Alzheimer's disease (AD).There are no effective treatments targeting NPS in AD.The current effort aims to collate, harmonize, and analyze all NPS data from the Alzheimer's Disease Sequencing Project.Core analyses will identify underlying genetic factors and mechanistic pathways.The harmonized genomic and phenotypic data from this initiative will be available through National Institute on Aging Genetics of Alzheimer's Disease Data Storage Site., Competing Interests: There are no competing interests to report. Author disclosures are available in the supporting information., (© 2024 The Author(s). Alzheimer's & Dementia: Diagnosis, Assessment & Disease Monitoring published by Wiley Periodicals LLC on behalf of Alzheimer's Association.)

Introduction: Plasma phosphorylated threonine-181 of Tau and amyloid beta are biomarkers for differential diagnosis and preclinical detection of Alzheimer disease (AD). Given differences in AD risk across diverse populations, generalizability of existing biomarker data is not assured., Methods: In 2,086 individuals of diverse genetic ancestries (African American, Caribbean Hispanic, and Peruvians) we measured plasma pTau-181 and Aβ42/Aβ40. Differences in biomarkers between cohorts and clinical diagnosis groups and the potential discriminative performance of the two biomarkers were assessed., Results: pTau-181 and Aβ42/Aβ40 were consistent across cohorts. Higher levels of pTau181 were associated with AD while Aβ42/Aβ40 had minimal differences. Correspondingly, pTau-181 had greater predictive value than Aβ42/Aβ40, however, the area under the curve differed between cohorts., Discussion: pTau-181 as a plasma biomarker for clinical AD is generalizable across genetic ancestries, but predictive value may differ. Combining genomic and biomarker data from diverse individuals will increase understanding of genetic risk and refine clinical diagnoses., Competing Interests: Conflict of Interest The authors declare no conflicts of interest.