Fig. 3. Differentially methylated positions (DMPs) identified in a cross-cortex meta-analysis include sites that are annotated to genes previously implicated in Alzheimer’s disease.

a Manhattan plot highlighting significant cortical DMPs associated with Braak NFT Stage from an EWAS meta-analysis of all available AD datasets (total N = 2013 individuals). In total 334 DMPs associated with tau pathology were identified using linear regression models controlling for major covariates (see Methods) at an experiment-wide significance threshold (P < 9E–08). The x-axis depicts individual chromosomes 1–22 and the y-axis gives the significance level (–log10(P)) for each DNA methylation site tested. The horizontal red line represents the experiment-wide significance threshold (P < 9E–08). Gene annotations are given for the 50 top-ranked DMPs and a full list of results is given in Supplementary Data 9. Many of the DMPs associated with tau pathology have been previously implicated in AD. Elevated tau pathology is associated with b hypermethylation at cg07061298 (effect size = 0.32%, SE = 0.037%, P = 8.06E–18) that is annotated to HOXA3 that has been implicated in previous EWAS analyses of AD pathology, c hypermethylation at cg05066959 (effect size = 0.41%, SE = 0.056%, P = 1.16E–13) that is annotated to ANK1 that is also strongly implicated in previous EWAS analyses of AD pathology, and d hypermethylation at cg06784824 (effect size = 0.21%, SE = 0.032%, P = 1.71E–11) that is annotated to SPI1 that is implicated in GWAS analyses of AD. The x-axis shows the effect size (% DNA methylation difference per SD increase in Braak NFT stage), with squares representing effect size and arms indicating the 95% confidence intervals. Details on each of the cohorts included in the meta-analysis (AZ1 Arizona 1, AZ2 Arizona 2, BDR Brains for Dementia Research, LBB1 London 1, LBB2 London 2, MS Mount Sinai, ROSMAP Religious Orders Study/Memory and Aging Project) are given in Supplementary Data 7.