15 Mar 2025

Screening large genetic datasets can uncover harbingers of neurodegenerative disorders such as Alzheimer’s disease. For AD, most older genome-wide association studies have relied heavily on participants with European ancestry, though evidence suggests at least all-cause dementia affects people of non-Hispanic black ancestry at a higher rate than those of non-Hispanic white ancestry. Is that due to genetics? “Lack of diverse genetic data is a major problem for AD GWAS studies,” said Dmitry Prokopenko and Rudy Tanzi of Massachusetts General Hospital, Charlestown, who led a large meta-analysis of whole-genome sequencing data incorporating datasets from people with diverse backgrounds. Their findings, published in the February 2025 Alzheimer’s & Dementia, point to 16 new loci with potential roles in AD pathogenesis.

First authors Julian Willett, Mohammad Waqas, and colleagues performed a two-phase analysis. They focused on cohorts with clinical AD diagnoses before broadening their inquiry to cover proxy AD cases, namely people who had a parent or grandparent diagnosed with the disease but no such diagnoses themselves. Their clinical AD data analysis began with a GWAS on a clinical AD cohort in which nearly half of participants had non-European ancestry, the dataset from the National Institute on Aging Genetics of Alzheimer’s Disease Data Storage Site (NIAGADS, see Dec 2013 community news). Then came a meta-analysis of the NIAGADS data together with clinical AD data from a family-based cohort from the National Institute of Mental Health.

Willett and colleagues found variants at 22 loci to associate with AD with genome-wide significance. Fourteen of them were new. Of these, nine rare variants lay close to the following genes: VWA5B1, RNU6-755P/LMX1A, MOB1A, MORC1-AS1, LINC00989, PDE4D, RNU2-49P/CDO1, NEO1, and SLC35G3/AC022916.1. Five common variants juxtaposed lncRNA AC090115.1/ZNF641, FBN2/ SLC27A6.28, DYM, KCNG1/ lncRNA AL121785.1, and TIAM1 genes. All but one the variants close to lncRNA AC090115.1/ZNF641 were protective for AD (image below).

Click to Enlarge

Variants in Clinical Alzheimer’s. Manhattan plot of NIAGADS/NIMH meta-analyses identified variants at 22 loci, 14 of them new (NL). [Image courtesy Willett et al., Alz & Dem., 2025.]

Armed with these results, the researchers expanded their study to two much larger cohorts: the U.K. Biobank and All of Us, a data repository established by the NIH in 2018 to promote inclusive health research. While more than 90 percent of U.K. Biobank participants had European ancestry, half of the people in All of Us did not. The AD status for most participants in these two cohorts was unknown, therefore the authors relied on AD-by-proxy, defining individuals who reported having a grandparent, parent, or sibling with AD as being at risk for the disease.

Analyzed this way, in the U.K. Biobank, 30 loci associated with AD, while 1,558 did in All of Us. The scientists validated these results using the NIAGADS/NIMH clinical data. Eighteen loci from the U.K. Biobank data had been associated with AD in prior analysis. In fact, 15 were near the APOE gene. From the All of Us cohort, only 10 loci replicated in the NIAGADS clinical AD datasets. A meta-analysis of both AD-by-proxy cohorts yielded 19 loci that were only nominally significant in the clinical data meta-analysis. Two new loci, both of them rare, came out of the by-proxy data, bringing the number of novel AD loci to 16. One lay near the genes RPL23 and LASP1; the former encodes a ribosomal protein whose expression is reportedly up in AD (Shigemizu et al., 2020). LASP1 encodes a protein found in synapses and dendritic spines linked to cognitive alterations in schizophrenia (Lin et al., 2019). 

Typically, GWAS hits are not coding variants, meaning scientists must probe further to identify associated functional genes. Willett, Waqas, and colleagues took a two-step approach. First, they turned to existing gene annotation databases, including FAVOR, Genehancer, and SuperEnhancer, to predict if any of the variants change gene expression by altering enhancer activity. Then they narrowed down that list to variants affecting genes up- or downregulated in AD, as predicted by prior single-cell RNA-Seq analysis (see Oct 2023 news on Mathys et al., 2023).

The clinical meta-analysis yielded 81 variants in 11 loci that might affect enhancers close to 28 genes differentially expressed in AD. For example, the enhancer-linked rs147450666 variant lives near the genes FBN2 and SLC27A6. The former ticks up in inhibitory neurons in AD, the latter does so in inhibitory and excitatory neurons. The proxy-AD analysis yielded two novel loci that might change expression of RPL23/LASP1 and CEBPA (table below).

Some genes were expressed differently based on a participant’s disease state. For instance, among people with neurofibrillary tangles, RPL23 expression in excitatory neurons was significantly higher in those with mild cognitive impairment than in those without.

Which Cells Express New Variants? Variants within novel loci link to genes that are differentially expressed in brain cells where there is evidence of ongoing AD pathology. Column on right lists affected cells, where Exr, Inh, Oli, Ast, and Mic correspond to excitatory neurons, inhibitory neurons, oligodendrocytes, astrocytes, and microglia, respectively. The comparators, e.g. 4v1, refer to (1) no pathology or cognitive impairment; (2) impairment but no pathology; (3) pathology but no impairment; (4) pathology and impairment. Numbers in parenthesis refer to the number of unique cell population comparisons. Plus and minus signs indicate log-fold change for each of those comparisons. [Courtesy of Willett et al., Alzheimer’s and Dementia, 2025.]

The use of the AD-by-proxy phenotype concerned Valentina Escott-Price of Cardiff University, who warned that this is prone to bias (see Wu et al., 2024). Prokopenko acknowledged problems with proxy GWAS but said their study mitigates them by validating analyses with findings from clinical AD data. The U.K. Biobank study asks participants about their family history of dementia, not AD, and Escott-Price thinks that might could skew the data. Prokopenko told Alzforum that because Alzheimer’s disease is a leading cause of dementia, the U.K. Biobank data fit within their AD-by-proxy framework.

Carlos Cruchaga of Washington University in St. Louis was surprised that the researchers didn’t replicate more of the 80-plus loci identified in previous AD GWAS work on datasets with mostly European individuals (see Feb 2021 news). Prokopenko said the statistical significance of many of these known loci fell once the authors corrected for population stratification, a phenomenon whereby different genetic variants tend to be more common in certain ancestral populations. Without correction, population stratification can produce false positives in multi-ancestry studies (Hellwege et al., 2017).

This research suggests there are “more genes and variants that are involved in Alzheimer’s disease compared to those that we have already identified as a community,” Cruchaga told Alzforum. “Additional studies are needed to fully replicate these novel findings.” Prokopenko and colleagues hope to use gene-based testing, combining multiple variants into a single test of disease association.—Lauren Schneider

Lauren Schneider is a freelance writer in New York City.

Comments

No Available Comments

Make a Comment

To make a comment you must login or register.

References

News Citations

1. Stunning Detail: Single-Cell Studies Chart Genomic Architecture of AD 7 Oct 2023
2. Massive GWAS Meta-Analysis Digs Up Trove of Alzheimer’s Genes 24 Feb 2021

Paper Citations

1. Shigemizu D, Mori T, Akiyama S, Higaki S, Watanabe H, Sakurai T, Niida S, Ozaki K. Identification of potential blood biomarkers for early diagnosis of Alzheimer's disease through RNA sequencing analysis. Alzheimers Res Ther. 2020 Jul 16;12(1):87. PubMed.
2. Lin CH, Yang S, Huang YJ, Lane HY. Polymorphism in the LASP1 gene promoter region alters cognitive functions of patients with schizophrenia. Sci Rep. 2019 Dec 11;9(1):18840. PubMed.
3. Mathys H, Peng Z, Boix CA, Victor MB, Leary N, Babu S, Abdelhady G, Jiang X, Ng AP, Ghafari K, Kunisky AK, Mantero J, Galani K, Lohia VN, Fortier GE, Lotfi Y, Ivey J, Brown HP, Patel PR, Chakraborty N, Beaudway JI, Imhoff EJ, Keeler CF, McChesney MM, Patel HH, Patel SP, Thai MT, Bennett DA, Kellis M, Tsai LH. Single-cell atlas reveals correlates of high cognitive function, dementia, and resilience to Alzheimer's disease pathology. Cell. 2023 Sep 28;186(20):4365-4385.e27. PubMed.
4. Wu Y, Sun Z, Zheng Q, Miao J, Dorn S, Mukherjee S, Fletcher JM, Lu Q. Pervasive biases in proxy genome-wide association studies based on parental history of Alzheimer's disease. Nat Genet. 2024 Dec;56(12):2696-2703. Epub 2024 Nov 4 PubMed.
5. Hellwege JN, Keaton JM, Giri A, Gao X, Velez Edwards DR, Edwards TL. Population Stratification in Genetic Association Studies. Curr Protoc Hum Genet. 2017 Oct 18;95:1.22.1-1.22.23. PubMed.

External Citations

1. FAVOR
2. Genehancer
3. SuperEnhancer
4. ).

Further Reading

News

1. Studying Diverse Populations May Require New Biomarkers 15 Feb 2025
2. Like GWAS, BWAS Need Thousands of Participants 18 Mar 2022