The largest Alzheimer’s GWAS yet has netted 42 new risk loci. The haul, described in a preprint on medRxiv, comes from an enormous collaborative effort by Alzheimer’s geneticists, with 354 authors. They combined findings from multiple European datasets, ultimately mining 111,326 cases and 677,663 controls to unearth 75 risk associations. Of these, 33 were known. In most of the 42 new loci, the scientists were able to pinpoint a likely causal gene. These fell into established pathways—amyloid and tau metabolism, endocytosis, innate immunity. Their protein products interacted with those of known AD genes. In addition, the investigators combined all 75 loci into a polygenic risk score, which accurately predicted progression to dementia in multiple observational cohorts.

  • A massive GWAS meta-analysis netted 42 new risk loci for AD.
  • Candidate genes in these loci fell into familiar pathways such as immunity and APP metabolism.
  • A polygenic risk score based on all known AD variants predicted progression to dementia with high accuracy.

“Because of the number of genes characterized, this is a turning point in genetics,” corresponding author Jean-Charles Lambert at the Institut Pasteur de Lille in France told Alzforum. Some of these data were previously presented at last year’s Alzheimer’s Association International Conference (Jul 2020 conference news). 

Carlos Cruchaga at Washington University in St. Louis called the work a tour de force. “The genes identified in these studies are instrumental for understanding the pathways implicated in disease, and will lead to drug targets,” he wrote to Alzforum (full comment below). He was not involved in the study.

Pile on the Genes. Manhattan plot shows 75 loci linked to AD that passed statistical muster (dotted red line). The 42 red ones are new. [Courtesy of Bellenguez et al.]

Meanwhile, a recent GWAS meta-analysis, described in the February 15 Nature Genetics, turned up three of the same new genes: TSPAN14, CCDC6, and NCK2, along with one additional one, SPRED2. That analysis also confirmed 30 known AD risk loci. Researchers led by Andrew Bassett at the Wellcome Sanger Institute, Cambridge, England, U.K., searched for causal variants in all 34 loci using fine-mapping techniques. Their methodology confirmed many AD genes but also turned up some surprises, nominating a new candidate.

Bigger Is Better
GWAS have been trending bigger for years. A meta-analysis in 2013 pulled together 17,000 cases and turned up 19 risk loci, 11 of them new (Oct 2013 news). Geneticists at the International Genomics of Alzheimer’s Project analyzed more than 35,000 cases to find five new loci (Mar 2019 news). A separate effort went broader by including UK Biobank samples from people with family histories of dementia, known as genome-wide association-by-proxy (GWAX); it uncovered nine new genes among almost 72,000 diagnosed and proxy cases (Apr 2018 news). 

For this study, geneticists took a hybrid GWAS/GWAX approach. Like the previous effort, they tapped genetic samples from the UK Biobank, but they combined these with numerous European GWAS, including the European Alzheimer’s Disease Biobank, the European Alzheimer’s Disease Initiative, GERAD/PERADES, the Spanish study GR@ACE, the Norwegian study DemGene, the Rotterdam study, and the Bonn study. Altogether, this meta-analysis brought together 39,106 clinically diagnosed AD cases, 46,828 proxy AD cases, and 401,577 controls for the discovery cohort. Lambert noted that about 30,000 of these cases and proxy cases had never been included in a GWAS before. He believes the number of new cases, as well as the size and power of the study, explains how it was able to find so many new disease associations.

The seven joint first authors are Céline Bellenguez at the Institut Pasteur de Lille; Fahri Küçükali at the University of Antwerp, Belgium; Iris Jansen at Vrije University, Amsterdam; Victor Andrade at the University of Cologne, Germany; Sonia Moreno-Grau at the Universitat Internacional de Catalunya in Barcelona, Spain; Najaf Amin at Erasmus MC in Rotterdam, the Netherlands; and Adam Naj at the University of Pennsylvania, Philadelphia. The scientists analyzed more than 21 million SNPs, imputing missing data using the Trans-Omics for Precision Medicine reference panel. This new panel is more comprehensive than previous haplotype references, being based on whole-genome sequencing of more than 90,000 people (Kowalski et al., 2019). 

“Usage of the TOPMed reference panel may have enhanced imputation accuracy, resulting in the identification of the new risk signals,” Nancy Ip at the Hong Kong University of Science and Technology wrote to Alzforum (full comment below). She was not involved in the study.

The authors tested the hits in a second large cohort of 25,392 AD cases and 276,086 controls from the combined ADGC, FinnGen, and CHARGE GWAS. In addition to confirming the 42 new loci, they also found 39 signals in 33 known AD loci that replicated in both cohorts.

Cellular Collaboration. Genes candidates in the 42 new loci are expressed by a variety of cell types, implicating cellular interactions in AD pathogenesis. [Courtesy of Bellenguez et al.]

Lambert notes that a number of regions of the genome seem to harbor more than one risk variant. However, researchers tend to focus on just a single candidate gene. “It likely means we are underestimating the impact of several loci, because we’re unable to capture the full genetic information in each locus,” he told Alzforum. Tracking down these multiple independent signals will be a crucial next step, he added.

The researchers attempted to home in on candidate genes in the new loci by chasing coding variants, eQTLs that control expression, splicing variants, or methylation changes. For 34 of the 42 new regions, they were able to nominate a single candidate gene. Many fall into the expected molecular pathways. Several are involved in APP processing; they include APP itself, the lysosomal gene CTSB, ADAM17, and TSPAN14, which regulates the known AD gene ADAM10, aka α-secretase. ADAM10 also turned up in this study. It shunts APP into a non-amyloidogenic pathway.

There were no tau-interacting proteins among the 42 new genome-wide significant hits. Intriguingly, numerous loci containing such genes had suggestive associations with AD. This list includes the kinase Fyn, the kinase GSK3β, microtubule-associated protein 2 (MAP2), and microtubule-affinity regulating kinase 4 (MARK4).

In agreement with other studies, lysosomal and endosomal genes cropped up repeatedly. The 42 new loci included the sorting receptors SORT1 and SNX1, the endosomal regulator WDR81, the lysosomal protease cathepsin H (CTSH), the lysosomal hydrolase IDUA, and the lysosomal trafficking regulator TMEM106B. Lipid metabolism made an appearance as well, with genes such as the transcriptional repressor JAZF1, which suppresses lipid production, and the cholesterol pump ABCA1.

As expected, immune-related genes represented one of the biggest categories. In particular, the study highlighted a role for the Linear Ubiquitination Assembly Complex. LUBAC inhibits TNF-α signaling, dousing inflammation. Two of its component proteins, RBCK1 and SHARPIN, showed up as AD-related, along with OTULIN, a de-ubiquitinase that regulates LUBAC. Other immune-related genes included TNIP1, RHOH, BLNK, SIGLEC11, LILRB2, LIME1, and progranulin.

Julie Williams at Cardiff University, Wales, U.K., a co-author, believes the 42 new associations likely represent genuine AD risk factors. “These are not random genes,” she told Alzforum. “They tell a coherent story that builds on previous publications that implicated these processes in AD.” In line with this, a recent study combined proteome-wide association data with GWAS results to identify 10 new AD genes. Four of these— ICA1L, PLEKHA1, DOC2A, and CTSH—also showed up among the 42 new loci in the meta-analysis, demonstrating that different methods are converging on the same genes and pathways (Feb 2021 news). 

Gene expression analyses showed the 42 new loci contained an unusual number of microglial genes, with 13 of the new candidates expressed exclusively or primarily in these cells. These included many of the immune genes, such as OTULIN, SHARPIN, RHOH, BLNK, SIGLEC11, LILRB2, and progranulin. But there were also microglial genes related to other processes, including JAZF1, TSPAN14, the adaptor protein NCK2, the lysosomal peptidase cathepsin B, the signaling protein RASGEF1C, and the transcription factor MAF. That said, candidate genes were expressed in many other brain cell types too. “There is clearly cross talk between cell types in AD,” Lambert noted.

Some of the new candidate AD genes associate with other neurodegenerative diseases as well, such as IDUA with Parkinson’s disease, and progranulin and TMEM106B with frontotemporal dementia. A co-localization analysis of the AD and PD signals determined that the AD risk variant near IDUA was probably independent of the PD signal. However, the variants near progranulin and TMEM106B likely contribute to risk for both disorders, the authors found. Cruchaga noted that other studies have implicated TMEM106B in AD (Li et al., 2020). 

After all this, how many AD genes remain to be discovered? This is unknown, researchers agreed. One recent study estimated there might be around 100 AD genes total, though other researchers dispute this (Zhang et al., 2020). “This would suggest that GWAS are potentially reaching the limit of discovering new loci associated with AD. However, our knowledge of the genetic architecture underlying AD is still far from complete,” Shea Andrews at Icahn School of Medicine at Mount Sinai, New York, wrote to Alzforum. He was not involved in the study.

What percentage of total AD genetic risk do the 75 known loci represent? This is also unknown, because case-control studies are not designed to answer this question. Researchers also do not know how much of the AD risk in a population is inherited, versus due to other factors such as lifestyle. Without knowing the heritability, geneticists can’t determine if they have found all variants that explain it. And since AD prevalence rises with age, heritability is hard to determine. “Estimates of heritability are extremely problematic in diseases with age-dependent penetrance,” John Hardy at University College London wrote to Alzforum (full comment below).

A related question is how well the known variants predict disease. On this, the scientists got more traction. They combined 83 independent variants from the 75 loci into a polygenic risk score and tested its predictive power in numerous observational studies, adjusting for age, sex, and APOE genotype. On average, the presence of one risk variant upped a person’s odds of developing AD by 5 percent. Multiple variants had a cumulative effect, such that a person who inherited 18 to 20 risk variants would have the same increased risk as someone with one APOE4 allele, or about 200 percent, Lambert told Alzforum.

This polygenic risk score was about 84 percent accurate in predicting progression, Williams told Alzforum. Lambert noted, “I was surprised it worked so well.” The variants had the same effect on progression to dementia in people with mild cognitive impairment as they did in healthy elderly people. In other words, the risk did not depend on disease stage.

Besides Lambert, senior authors on this study were Mikko Hiltunen at the University of Eastern Finland in Kuopio, Kristel Sleegers at UAntwerp, Gerard Schellenberg at UPenn, Cornelia van Duijn at the Nuffield Department of Population Health at Oxford University, U.K., Rebecca Sims at Cardiff, Wiesje van der Flier at Amsterdam University, Agustín Ruiz at Universitat Internacional de Catalunya, and Alfredo Ramirez at University of Cologne.

What’s Next?
Lambert believes GWAS can grow bigger yet. He’d like to combine all known datasets into one massive meta-analysis for maximum power. Even so, Williams noted that GWAS overlook rare and structural variants. For those, geneticists need additional familial studies and specialized techniques such as long-read sequencing, which enable them to detect insertions and deletions.

Nab Those Causal Variants. At 37 loci linked to AD, an analysis of protein-coding changes, co-localization with eQTLs, distance to nearby genes, and interaction with AD gene networks spat out putative causal genes. Model score shows the likelihood of each gene being causal. [Courtesy of Schwartzentruber et al., Nature Genetics.]

For their part, Bassett and colleagues performed a smaller meta-analysis, combining data from a previous UK Biobank GWAX with the latest GWAS data from IGAP (Jansen et al., 2019; Kunkle et al., 2019). This identified 34 genome-wide risk loci, four of them new. The one that did not appear in the Bellenguez study, SPRED2, regulates the MAP kinase cascade.

First author Jeremy Schwartzentruber made use of numerous lines of evidence to nominate likely causal variants and genes in each of the 34 loci, plus three with suggestive associations to AD. For each SNP, Schwartzentruber and colleagues considered its distance to nearby genes, whether it abutted an eQTL for that gene, and whether the gene was connected to known AD genes through protein-interaction networks. For eQTL data, they used multiple samples and cell types, including primary microglia, blood immune cells, and different brain cortical samples. The authors also fine-mapped each locus using data on open chromatin structure to find variants likely to affect gene expression.

“This is impressive work,” Williams noted. “Using a variety of tissues was novel and very informative.”

For many AD loci, such as BIN1, PICALM, and SORL1, the analysis confirmed the likely involvement of that gene. For other loci, however, the authors nominated new candidates. In the presumed ADAMTS4 locus, they suggested the microglial gene FCER1G might be the causal factor instead, and for the EPHA1 locus, the microglial gene zyxin. In the SLC24A4 locus, they found evidence for endocytic gene RIN3 being the functional variant, as have other recent studies (Aug 2017 news; Aug 2019 news; Nov 2019 news). In the CLNK locus, they proposed HS3ST1, a heparan sulfate sulfotransferase, which facilitates uptake of tau by cultured cells and was recently associated with AD (Zhao et al., 2020). 

Overall, larger GWAS and multiple methods appear to be homing in on the same large set of genes involved in AD. “We are obtaining a full picture of locations in the genome that may harbor important genetic variants that would facilitate the ability to identify potential targets for drug development,” Richard Mayeux at Columbia University Medical Center in New York wrote to Alzforum (full comment below). He is a co-author on the Bellenguez paper.—Madolyn Bowman Rogers

Comments

  1. It is great to see new putative genetic risk loci.

    Two issues: Are these new genetic risk loci certain? And how much of the genetic risk has been found?

    On the first point: The power of “simple GWAS” is that the statistical test is easy to understand. Its weakness is that the multiple test correction is so stringent that many true positives are lost. Any process that deviates from the simple GWAS by incorporating other factors (such as co-expression or biological information) undoubtedly helps find new loci, but are they certainly correct? I would not say “certain” but would say “likely.” Cheating Bonnferoni is good. … But comes at a price.

    One the second point of “missing heritability” and how much has been found: The problem is not how much has been found (the numerator), but rather how much there is to find (the denominator). Estimates of heritability are extremely problematic in diseases with age-dependent penetrance, especially if one believes (as I do) that we would all get it if we lived long enough.

  2. Undertaking genetic studies to identify causal factors for Alzheimer’s disease (AD) requires sufficient sample size and detailed analysis. In this study led by Bellenguez and Lambert et al., genome-wide association analysis (GWAS) based on imputed array data with a total of 788,989 participants, have identified 75 AD risk loci, 42 of which are novel AD risk loci.

    It is quite promising that new AD risk loci can be derived from this GWAS analysis. One possible reason for the new finding may be due to the utilization of a larger sample size in this study compared to previous studies (Schwartzentruber et al., 2021; Kunkle et al., 2019; Marioni et al., 2018). Moreover, instead of a commonly used imputation panel such as the 1000 Genomes Project data or the Haplotype Reference Consortium (HRC) reference panel, the authors used the Trans-Omics for Precision Medicine (TOPMed) reference panel, for the imputation of the array data used in this study. This is a newly launched panel comprising deep whole-genome sequencing (WGS) data (~30X) from more than 90,000 individuals and covering more than 300 million genomic variants (Kowalski et al., 2019; Taliun et al., 2021). Notably, it has been demonstrated that the selection of the reference panel for genotype imputation is critical for the discovery of disease-associated risk variants (McCarthy et al., 2016). 

    Thus, usage of the TOPMed reference panel may have enhanced imputation accuracy, resulting in the identification of the new risk signals. In addition, with the inclusion of more variants in the reference genome panel, this study has enabled a better recovery of rare variants or haplotype structures that are associated with AD (such as NCK2 rs143080277, identified in this study).

    With the identification of 42 new gene loci in this study, the authors then conducted fine-mapping analysis to identify key AD risk genes in each novel locus (e.g., EGFR in locus 18). This opens a new horizon for identifying candidate risk genes for subsequent mechanistic studies.

    However, the current imputed array approach also has some limitations: The haplotype structures could not be recovered by the limited variants assayed in the array, and the disease causal variants may not be captured by the array due to the limited detection capacity of the array genotyping. Thus, WGS, which can capture most of the genomic variations in the studied participants, would be a better option for both GWAS as well as fine mapping analysis of a specific locus (Prokopenko et al., 2020; Zhou et al., 2019). 

    Corroborating other studies, this work highlights the involvement of the immune pathway in AD pathogenesis. Thus, the identification of the novel loci may lead to the discovery of new molecular pathways in the immune system. Furthermore, as our previous AD GWAS study in the Chinese population also identified risk candidates that associate with immune functions (Zhou et al., 2018), is important to investigate whether and how ethnic backgrounds influence immune or other AD-associated pathways at the genetic level.

    References:

    . Genome-wide meta-analysis, fine-mapping and integrative prioritization implicate new Alzheimer's disease risk genes. Nat Genet. 2021 Mar;53(3):392-402. Epub 2021 Feb 15 PubMed.

    . Genetic meta-analysis of diagnosed Alzheimer's disease identifies new risk loci and implicates Aβ, tau, immunity and lipid processing. Nat Genet. 2019 Mar;51(3):414-430. Epub 2019 Feb 28 PubMed.

    . GWAS on family history of Alzheimer's disease. Transl Psychiatry. 2018 May 18;8(1):99. PubMed.

    . Use of >100,000 NHLBI Trans-Omics for Precision Medicine (TOPMed) Consortium whole genome sequences improves imputation quality and detection of rare variant associations in admixed African and Hispanic/Latino populations. PLoS Genet. 2019 Dec;15(12):e1008500. Epub 2019 Dec 23 PubMed.

    . Sequencing of 53,831 diverse genomes from the NHLBI TOPMed Program. Nature. 2021 Feb;590(7845):290-299. Epub 2021 Feb 10 PubMed.

    . A reference panel of 64,976 haplotypes for genotype imputation. Nat Genet. 2016 Oct;48(10):1279-83. Epub 2016 Aug 22 PubMed.

    . Whole-genome sequencing reveals new Alzheimer's disease-associated rare variants in loci related to synaptic function and neuronal development. medRxiv. 2020 Nov 4; PubMed.

    . Non-coding variability at the APOE locus contributes to the Alzheimer's risk. Nat Commun. 2019 Jul 25;10(1):3310. PubMed.

    . Identification of genetic risk factors in the Chinese population implicates a role of immune system in Alzheimer's disease pathogenesis. Proc Natl Acad Sci U S A. 2018 Feb 20;115(8):1697-1706. Epub 2018 Feb 5 PubMed.

  3. This work is clearly a tour de force, but also a good example of what can be achieved through broad collaboration.

    It is worth noting that identifying genes or genetic regions just for the sake of it is useless. The important aspect is that the genes identified in these studies will be instrumental in understanding the pathways implicated in disease. This paper offers a good example because genes enriched in the endolysosomal pathway specially emerged. Other pathways involving microglia (TREM2 and others) or Ab metabolism are also very important and this study confirms their roles in AD.

    The authors were able to find so many new gene candidates because of the new AD cohorts they used and because of the UK Biobank data, which is increasing the number of samples significantly. However, it must be kept in mind that the UK Biobank phenotypes are not as clean as those in AD-specific studies. The biobank covers dementia in general, as the authors reflect in the title and the manuscript.

    This may explain why some hits are difficult to intercept, such as TMEM106B and GRN, which are FTD genes, and IDUA, which is a PD signal. These associations could indicate some contamination with FTD or PD samples in the UK Biobank, or that some common pathways exist between AD and FTD, or AD and PD. Previously, several studies, including some from us, suggested that TMEM106B is involved in AD (Li et al., 2020Yang et al., 2020). 

    There are still a lot of genes to find. If we compare with other diseases, such as PD or psychiatry disorders, we can expect to find many new AD genes with additional GWAS, but also with sequencing studies. More GWAS hits will allow us to create better prediction models, lead to more pathways, and also identify new drug targets. Right now, there are candidate drugs or even trials targeting TREM2, CD33, MS4A4a and SPI1.

    But there is still a lot of work to do. For some loci we do not yet know the functional gene, so combining this data with novel approaches such as co-localization or Mendelian randomization will be necessary.  

    References:

    . The TMEM106B FTLD-protective variant, rs1990621, is also associated with increased neuronal proportion. Acta Neuropathol. 2020 Jan;139(1):45-61. Epub 2019 Aug 27 PubMed.

    . Genetics of Gene Expression in the Aging Human Brain Reveal TDP-43 Proteinopathy Pathophysiology. Neuron. 2020 Aug 5;107(3):496-508.e6. Epub 2020 Jun 10 PubMed.

  4. GWAS finds regions in the genome, not genes per se. The difficult part is actually finding the underlying causal variant. However, we are obtaining a full picture of locations in the genome that may harbor important genetic variants that would facilitate the ability to identify potential targets for drug development.

    Data from the UK Biobank lacks diagnostic precision. Afterall, having only a family history of Alzheimer’s disease can make one a “case” in this cohort.  It is also clear that the misdiagnosis of Alzheimer’s disease, clinically, is somewhere between 10 percent to 20 percent. Therefore, it is possible that some of the loci reported are not related to Alzheimer’s disease, but some other form of dementia. 

  5. The new genome-wide association study by Bellenguez et al. identified 75 risk loci associated with AD, of which 42 were novel. This effort was able to discover so many new loci primarily by increasing the sample size of the study, with the number of clinical cases increasing to 57,169 from 35,274 (Kunkle et al., 2019; Mar 2019 news). In addition, they included 46,828 “proxy cases” from the UK Biobank, where individuals reported that either one or both of their biological parents had dementia. Furthermore, by using the TOPMed imputation panel they were able to increase the number of the variants tested (21 million) and the imputation quality of rare variants.

    In addition to conducting a GWAS to identify risk loci, the authors also performed a series of analyses integrating molecular quantitative trait loci (QTL) datasets to prioritize candidate causal genes at each locus. This analysis proposed five new candidate genes (DGKQ, RASA1, ICA1, DOC2A, and LIME1) that modulate APP metabolism and nine new candidate genes that are highly expressed in microglia (OTULIN, RASGEF1C, TSPAN14, BLNK, ATP8B4, MAF, GRN, SIGLE11C and LILRB2). Taking into account genes that were previously linked to microglia function from earlier GWAS, 25 percent of the loci described by Bellenguez and colleagues are credibly linked to AD-related microglia dysfunction. Interestingly, the authors also observed a statistical interaction between APP/Aβ pathways and expression of genes in microglia, suggesting that there is a functional relationship between these two pathways.

    It has recently been suggested that there may only be ~100 common causal variants associated with risk of AD—though there is some debate around this (Zhang et al., 2020, and related comments). This would suggest that GWAS are potentially reaching the limit of discovering new loci associated with AD. However, our knowledge of the genetic architecture underlying AD is still far from complete.

    To date, most GWAS of AD have been conducted in populations of European ancestry. As such it is imperative to expand genetic studies of AD into other underrepresented ethnic groups, which will likely highlight new risk loci and improve the mapping of functional variants. Furthermore, ongoing whole-exome and whole-genome sequencing studies will likely further identify rare genetic variation that are not well captured by GWAS.

    Finally, GWAS of other AD endophenotypes (CSF Aβ/tau biomarkers, neuropathology, progression, resilience), which are more proximal to the direct effect of a gene, will identify novel loci and further aid in the prioritization of likely causal genes.

    References:

    . Genetic meta-analysis of diagnosed Alzheimer's disease identifies new risk loci and implicates Aβ, tau, immunity and lipid processing. Nat Genet. 2019 Mar;51(3):414-430. Epub 2019 Feb 28 PubMed.

    . Risk prediction of late-onset Alzheimer's disease implies an oligogenic architecture. Nat Commun. 2020 Sep 23;11(1):4799. PubMed.

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. Doubling Down on Sequencing Serves up More Alzheimer’s Genes
  2. Paper Alert: New Alzheimer’s Genes Published
  3. Paper Alerts: Massive GWAS Studies Published
  4. GWAS, GWAX: bioRχiv Hosts Bonanza of Alzheimer’s Genetics
  5. PWAS x GWAS? Proteome Analysis Nets 10 New Alzheimer’s Genes
  6. The Search for the Missing AD Heritability Turns Up New Rare Variants
  7. AD Genetic Risk Tied to Changes in Microglial Gene Expression
  8. Single-Cell Expression Atlas Charts Changes in Alzheimer’s Entorhinal Cortex

Paper Citations

  1. . Use of >100,000 NHLBI Trans-Omics for Precision Medicine (TOPMed) Consortium whole genome sequences improves imputation quality and detection of rare variant associations in admixed African and Hispanic/Latino populations. PLoS Genet. 2019 Dec;15(12):e1008500. Epub 2019 Dec 23 PubMed.
  2. . The TMEM106B FTLD-protective variant, rs1990621, is also associated with increased neuronal proportion. Acta Neuropathol. 2020 Jan;139(1):45-61. Epub 2019 Aug 27 PubMed.
  3. . Risk prediction of late-onset Alzheimer's disease implies an oligogenic architecture. Nat Commun. 2020 Sep 23;11(1):4799. PubMed.
  4. . Genome-wide meta-analysis identifies new loci and functional pathways influencing Alzheimer's disease risk. Nat Genet. 2019 Mar;51(3):404-413. Epub 2019 Jan 7 PubMed.
  5. . Genetic meta-analysis of diagnosed Alzheimer's disease identifies new risk loci and implicates Aβ, tau, immunity and lipid processing. Nat Genet. 2019 Mar;51(3):414-430. Epub 2019 Feb 28 PubMed.
  6. . 3-O-Sulfation of Heparan Sulfate Enhances Tau Interaction and Cellular Uptake. Angew Chem Int Ed Engl. 2020 Jan 27;59(5):1818-1827. Epub 2019 Dec 10 PubMed.

External Citations

  1. preprint
  2. GERAD/PERADES
  3. GR@ACE
  4. Rotterdam
  5. Bonn
  6. Trans-Omics for Precision Medicine

Further Reading

Primary Papers

  1. . New insights on the genetic etiology of Alzheimer’s and related dementia. medRxiv. 2020 Dec 14.
  2. . Genome-wide meta-analysis, fine-mapping and integrative prioritization implicate new Alzheimer's disease risk genes. Nat Genet. 2021 Mar;53(3):392-402. Epub 2021 Feb 15 PubMed.