Amyloid, tau and risk of Alzheimer's disease: a Mendelian randomization study

This study was carried out to assess the effect of amyloid and tau on Alzheimer's disease using two-sample Mendelian randomization design. Genetic associations with plasma amyloid species (amyloid precursor protein, amyloid-like protein 2, serum amyloid P-component, amyloid beta peptide), cerebrospinal fluid (CSF) amyloid beta, total tau, and phosphorylated tau₁₈₁ were extracted from the largest genome-wide association study (GWAS) available. Genetic associations with Alzheimer's disease were obtained from a GWAS of proxy-cases based on family history of Alzheimer's disease with 314,278 participants from the UK Biobank and a GWAS with clinically diagnosed Alzheimer's disease from the International Genomics of Alzheimer's Project (IGAP) with 21,982 cases and 41,944 controls. Estimates were obtained using inverse variance weighting with sensitivity analyses including MR-Egger, weighted median and MR-PRESSO. Presence of bias due to selective survival and competing risk was also considered. Plasma amyloid species, CSF total tau and phosphorylated tau₁₈₁ were not associated with Alzheimer's disease. For CSF Aβ₄₂, no association was found using the proxy-cases but an inverse association was found after removing outliers with MR-PRESSO using IGAP. Higher genetically predicted (p < 1 × 10^-5) plasma amyloid species, CSF total tau and phosphorylated tau₁₈₁ (based on sample sizes ~ 3300) were not associated with Alzheimer's disease using family history or clinically diagnosed cases while effects of CSF Aβ₄₂ were inconsistent between the family history and IGAP GWAS.