Consensus meta‐analysis of genome‐wide association studies for Alzheimer’s disease across the EADB, PGC‐ALZ and IGAP consortia
Céline Bellenguez- Psychiatry and Mental health
- Cellular and Molecular Neuroscience
- Geriatrics and Gerontology
- Neurology (clinical)
- Developmental Neuroscience
- Health Policy
- Epidemiology
Abstract
Background
A large number of putative genes have been associated with Alzheimer’s disease (AD), in particular by the International Genomics of Alzheimer’s Project (IGAP), the European Alzheimer & Dementia Biobank (EADB), and the Psychiatric Genomics Consortium (PGC). Those genome‐wide association studies (GWAS) showed that increasing the sample size and performing imputation using the largest reference panel available greatly improves statistical power to detect genetic associations with AD. In order to better characterize the genetic architecture and pathophysiology underlying this disease, the three consortia joined to perform a meta‐analysis of GWAS across all their samples of European ancestry, the UK Biobank, and FinnGen.
Method
In a preliminary analysis, 48 studies were considered, totaling 70,276 AD cases, 610,828 controls, 55,960 proxy‐cases and 235,566 proxy‐controls. This represents an effective sample size of 225,605 (97% of the expected final size). Most of the datasets were imputed with the TOPMed reference panel. A GWAS was performed in each study with a logistic regression or a logistic mixed model adjusted on principal components and/or centers, and results were combined across studies with an inverse variance fixed effect meta‐analysis.
Result
More than 23 million variants were analyzed. We identified a genome‐wide significant signal for 92 loci, including 16 novel loci. Lead variants were InDels for two novel loci, and had a frequency below 5% for four other novel loci. We did not identify any genome‐wide significant signals in six previously reported loci: results are borderline significant for SORT1 and ICA1, and further investigations are required for FOXF1, SLC2A4RG, HAVCR2 and NTN5.
Conclusion
This meta‐analysis will provide consensus results for loci previously reported for AD and will identify novel signals. Their robustness will be assessed after exclusion of proxy or biobank samples, and secondary analyses will be performed, such as pathway or functional analyses, to provide further insights into the novel loci.