![[Adobe Stock]](https://sp-ao.shortpixel.ai/client/to_webp,q_glossy,ret_img,w_750,h_375/https://www.psypost.org/wp-content/uploads/2026/08/dna-750x375.jpg)
An international genetic study of more than 2.8 million people has identified 127 regions of human DNA associated with Alzheimer’s disease, including 48 regions linked to the condition for the first time.
Published in Nature, the research indicates that common inherited variants account for nearly three times as much of the population variation in Alzheimer’s risk as previous estimates suggested. The findings also connect genetic susceptibility to heightened gene activity in brain immune cells alongside dampened gene activity in nerve cells.
Searching for Missing Genetic Risk in Alzheimer’s Disease
Alzheimer’s disease is a progressive brain disorder that gradually erodes memory, reasoning, and everyday functioning, accounting for most dementia cases worldwide. Existing disease-modifying drugs mainly aim to remove sticky clumps of amyloid-beta protein from brain tissue, yet they tend to offer modest symptom relief. Because twin studies indicate that inherited differences account for 60% to 80% of the variation in Alzheimer’s risk, researchers use genome-wide association studies, scanning millions of common single-letter DNA variations across populations to find chromosomal regions, called loci, linked to the illness.
Larger samples have steadily expanded that map. For example, a 2021 genetic study of more than 1.1 million people was led by several of the same researchers, including co-lead author Douglas Wightman, and shared several cohorts with the new work. That analysis uncovered seven new risk regions and highlighted microglia, the specialized immune cells in the brain and spinal cord that engulf debris and respond to injury. Soon after, a 2022 genome-wide study of nearly 790,000 individuals, which also shared several co-authors and datasets with the current study, mapped 75 risk regions and brought the total number of known Alzheimer’s loci to roughly 80.
Don’t miss the next big psychology finding
Every morning, PsyPost sends a short briefing on the most interesting new peer-reviewed research in psychology and neuroscience.
Even with dozens of risk regions mapped, statistical tallies of heritability suggested that common DNA variants explained only about 5% of population differences in risk, fueling doubts that much genetic risk remained to be found beyond APOE, a major risk gene that transports cholesterol and fats in the brain. The authors of a 2020 genetic modeling study argued that late-onset Alzheimer’s might involve fewer than 100 common causal variants in total, implying gene discovery had nearly plateaued.
Other models predicted roughly 1,200 contributing regions, prompting study co-lead authors Emil Uffelmann and Douglas P. Wightman, researchers in the Complex Trait Genetics lab at Vrije Universiteit Amsterdam, and their colleagues to test how much common genetic risk remained hidden.
“The main goal of our study was locus discovery: identifying variants and genes that affect the risk of developing Alzheimer’s disease,” Uffelmann told PsyPost. “Each new discovery provides clues about the causes of the disease and may lead to new treatments in the future. Another, more academic, motivating factor was to address the conundrum of low heritability estimates.”
Mapping 127 Genetic Regions Across 2.8 Million People
The team pooled data across 30 cohorts totaling more than 2.8 million participants: 109,479 diagnosed Alzheimer’s cases, 74,141 proxy cases, 2,131,799 controls, and 499,708 proxy controls. Because Alzheimer’s typically emerges after age 65, many biobank volunteers are too young to show symptoms, so genotyped adults who reported a parent with Alzheimer’s or dementia served as proxy cases, weighted to reflect the half of parental DNA that children inherit.
Measured by effective sample size, which weights cohorts by their balance of cases and controls, the dataset was 91% European, 5% African, 3% East Asian, and under 1% each of admixed American and South Asian ancestry.
Across the cohorts, the team analyzed more than 8.1 million common DNA variants, adjusting for age, sex, genetic ancestry, and study-specific factors while excluding cases diagnosed before age 40 where possible to avoid rare single-gene forms of the disease. Researchers then combined the genetic map with machine-learning tools, single-cell brain RNA atlases, and regulatory drug databases to identify likely causal genes, relevant cell types, and drug interactions. Analysis code and summary statistics, excluding proprietary 23andMe data, were posted publicly.
Across all meta-analyses, the researchers identified 127 distinct genetic regions associated with Alzheimer’s disease, consisting of 118 in the multi-ancestry analysis and nine additional regions in the European-only analysis. Forty-eight regions were newly identified, while associations on the X chromosome, tested in a smaller effective sample of 196,822 compared with 416,493 overall, were not statistically significant. Ancestry-specific scans detected 114 regions in European, three in African, and two in East Asian groups, whereas associations in the smaller South Asian and admixed American samples were not statistically significant.
A subset with direct clinical diagnoses rather than parental proxies included 20,402 female cases and 15,689 male cases, with 261,889 and 214,956 controls respectively. Inherited risk overlapped closely between men and women. However, the high-risk ε4 variant of APOE carried a modestly stronger association with Alzheimer’s disease in women than in men, about 13% stronger.
Microglial Overactivity and Neuronal Dampening
Comparing the genetic findings against single-cell RNA data from donated brain tissue showed that genetic risk concentrated in genes expressed at higher levels in microglia and lower levels in neurons, the nerve cells that transmit electrical and chemical signals across the brain, in people with Alzheimer’s compared with unaffected donors. Three neuronal subtypes initially stood out, including two types of inhibitory, or signal-braking, neurons that produce somatostatin or small signaling peptides and are known to be vulnerable early in Alzheimer’s. After adjusting for average expression changes across all brain cell types, however, the microglial link held firm while the individual subtype signals weakened, indicating broader neuronal dampening.
The authors suggest that inherited risk variants may push microglial immune genes toward overactivity while lowering gene expression in vulnerable neurons, though they note it remains uncertain whether neuronal dampening is an initial trigger or a downstream consequence of pathology. Pathway analyses reinforced roles for immune activation, fat binding, cell adhesion, and amyloid clearance, alongside five processes new to Alzheimer’s genetic scans, including oxidative-stress regulation via glutathione metabolism and membrane recycling at nerve junctions.
Higher Heritability and Sharper Risk Prediction
Using a statistical model that accommodates APOE’s large impact alongside functional DNA features, the team estimated that common variants account for 16% of population differences in underlying susceptibility to Alzheimer’s among people of European ancestry, assuming a 5% lifetime prevalence. That heritability estimate is nearly triple the 6% figure from an older regression method that simulations indicated understates heritability when one gene dominates, while African-ancestry heritability estimates were not statistically significant given smaller samples and greater genetic diversity.
Combined into a polygenic score, a single number summarizing inherited risk across millions of DNA variants, the predictors explained nearly 13% of the variation in disease susceptibility on average across held-out European cohorts, outperforming APOE status alone. Among European-ancestry UK Biobank adults aged 65 and older, Alzheimer’s prevalence was seven times higher in the top 10% of polygenic scores than in the bottom 10%, and three times higher than the whole-sample average.
Uffelmann offered a sharper version of the same comparison: “Thinking about genetic risk in absolute terms may be useful. We found that, on average, 1 in 20 participants aged 65 years or older in the UK Biobank were diagnosed with Alzheimer’s disease. In people with the highest genetic risk (top 1%), the fraction of Alzheimer’s cases increased to almost 1 in 4. Clearly, genetics plays an important role in affecting risk, but it does not determine diagnosis.”
Potential Drug Targets and Shared Brain Pathways
Cross-referencing newly prioritized genes with regulatory databases identified 17 genes in the new regions that interact with approved non-Alzheimer’s medications. Compounds targeting three of those genes showed promise in earlier animal studies of Alzheimer’s, whereas drugs targeting two others failed to ease symptoms in prior mouse or human tests.
Within a known risk region, the team’s model newly prioritized QPCT, which encodes an enzyme that modifies amyloid-beta into the specific form targeted by the approved Alzheimer’s antibody drug donanemab. A newly identified region spanning UBE2V1 and SPATA2, genes involved in protein clumping and programmed cell death, had previously been linked to both amyotrophic lateral sclerosis and frontotemporal dementia, and shared a likely causal variant between Alzheimer’s disease and amyotrophic lateral sclerosis.
These results align with an earlier multi-ancestry genetic analysis of clinically diagnosed and family-history cases, which uncovered 16 novel Alzheimer’s risk regions and linked several to altered brain gene expression. The cellular pattern also echoes an earlier post-mortem brain tissue study. Although that work measured active gene expression after death rather than inherited DNA variants across populations, it similarly pointed to heightened inflammatory activity in microglia paired with reduced signaling gene expression in neurons in Alzheimer’s disease.
Caveats and the Bigger Health Picture
There are a few things to keep in mind when interpreting these findings. Although public and nonprofit grants primarily funded the work, several co-authors are employed by or hold stock in pharmaceutical and biotechnology companies, including Roche, Regeneron, Amgen deCODE genetics, Janssen, Bayer, and Eli Lilly, which makes donanemab. Co-author Kaj Blennow became an Eli Lilly employee in September 2025. Several other authors have served as paid consultants or advisors to drugmakers developing Alzheimer’s treatments.
More than 90% of the effective sample also came from people of European descent, which limited locus discovery in underrepresented groups and made polygenic scores far less accurate in African and East Asian cohorts. Roughly 40% of all cases, and 62% to 82% in non-European groups, relied on self-reported parental dementia history rather than clinical diagnoses, potentially blending Alzheimer’s with other late-life dementias. Finally, a genetic link or drug-database match does not mean targeting a gene will safely slow cognitive decline, and polygenic scores remain probabilistic estimates rather than diagnoses, pointing to the need for laboratory experiments and more diverse clinical cohorts.
Stepping back, Uffelmann emphasized that inherited DNA is one piece of a broader health picture. “Genetics plays an important role in Alzheimer’s risk and points mostly to immune involvement, chiefly through microglia, the brain’s resident immune cells,” he said. “Yet it’s only part of the story, and lifestyle and environmental factors also matter.”
The study, “Alzheimer’s disease genetics converges on microglial and neuronal dysregulation,” was authored by Emil Uffelmann, Douglas P. Wightman, Shahram Bahrami, Alexey A. Shadrin, Vera Fominykh, Takafumi Ojima, Chenyang Jiang, Christian Benner, Elisa Moreno, Adrian I. Campos, Jesper Q. Thomassen, Emmanuel Minois-Genin, Hei Man Wu, G. Bragi Walters, Richard Sherva, Tian Lin, Xuemin Wang, Julien Bryois, Kristi Krebs, Marijn Schipper, Akira Narita, Alessandro Serretti, Anja H. Simonsen, Anna L. van Seumeren, Anne Corbett, Anne-Brita Knapskog, Annette M. Hartmann, Anouk den Braber, Argonde C. van Harten, Arvid Harder, Arvid Rongve, Bengt O. Madsen, Betty M. Tijms, Bitten Aagaard, Bjørn Lichtwarck, Bjørn-Eivind Kirsebom, Byron Creese, Chandra A. Reynolds, Sara Hägg, Ida K. Karlsson, Christian Erikstrup, Christina Mikkelsen, Clive Ballard, Dag Aarsland, Daichi Shigemizu, Dan Rujescu, Daniel Gudbjartsson, Eivind Aakhus, Erik Sørensen, Eystein Stordal, Flora H. Duits, Frank J. Wolters, Frederic Blanc, Geert Jan Biessels, Geir Selbæk, Geir Bråthen, Gen Tamiya, Gunhild Waldemar, Harro Seelaar, Helga Eyjolfsdottir, Henne Holstege, Henning Bundgaard, Henrik Zetterberg, Henrik Ullum, Ina Giegling, Ingmar Skoog, Ingrid T. Medbøen, Ingvild Saltvedt, Irena Rektorova, J. Michael Gaziano, Jan Haavik, Jens Hjerling-Leffler, Jiao Luo, Jon Snaedal, Everard G. B. Vijverberg, Julia M. Sealock, Kaj Blennow, Kaja Nordengen, Karin Persson, Katja Scheffler, Koichi Matsuda, Kouichi Ozaki, Lasse Pihlstrøm, Lavinia Athanasiu, Lene Pålhaugen, Marc Hulsman, Margda Waern, Maria Averina, Marianne Wettergreen, Marta R. Moksnes, Martijn Huisman, Masayuki Yamamoto, Mathias Toft, Matthew S. Panizzon, Mie Topholm Bruun, Mohsen Ghanbari, Monique Franc, Nancy L. Pedersen, Nathaniel Y. Bell, Niccoló Tesi, Ole B. Pedersen, Oleksandr Frei, Olivier Bousiges, Per Svenningsson, Pieter J. Visser, Qingqin S. Li, Richard Hauger, Rui Zhang, Shinichi Namba, Sigrid B. Sando, Silke Kern, Srdjan Djurovic, Steinunn Thordardottir, Tanya N. Phung, Thomas Truelsen, Thomas Werge, Thomas F. Hansen, Tomoki Kyosaka, Torgeir Engstad, Tormod Fladby, Victoria Merritt, Sverre Bergh, Wiesje M. van der Flier, Rujin Wang, Eli A. Stahl, Basavaraj Hooli, LifeLines Cohort Study, DBDS Genomic consortium, Regeneron Genetics Center, Penn Medicine Biobank, GHS-RGC DiscovEHR collaboration, Mayo Clinic-RGC Project Generation, Colorado Center for Personalized Medicine-RGC Collaboration, UCLA-RGC ATLAS collaboration, INDIANA-CHALASANI, Mount Sinai Million Health Discoveries Program, Estonian Biobank research team, VA Million Veteran Program, Lea K. Davis, Mark W. Logue, Kelli Lehto, Anna Zettergren, Ben M. Brumpton, Jian Zeng, Peter M. Visscher, Paul F. O’Reilly, Anubha Mahajan, Manuel Ferreira, Yukinori Okada, Sven J. van der Lee, Sisse R. Ostrowski, Ruth Frikke-Schmidt, Hreinn Stefansson, Karl Heilbron, Ole A. Andreassen, and Danielle Posthuma on behalf of the Alzheimer’s Disease Working Group of the Psychiatric Genomics Consortium.
