Investigation of potential causes of Alzheimer's disease using mendelian randomization

Abstract

Alzheimer’s disease and other dementias were the 7th leading cause of death globally in 2019 causing more than 1.5 million deaths. Global prevalence of Alzheimer’s disease is projected to reach 100 million by 2050. Despite decades of research, the causes of Alzheimer’s disease remain unclear. The amyloid hypothesis has been the main research focus for the past 2 decades, but amyloid targeting therapies or drugs have not shown definite effects in randomized controlled trials. Whether other hypotheses, such as the tau or inflammation hypotheses, explain the cause of Alzheimer’s disease is unknown. Previous evidence is mainly based on observational studies which are susceptible to confounding and selection bias, hindering the search for the possible causes. In order to examine less confounded associations of potential causes with risk of Alzheimer’s disease, Mendelian randomization analyses was used to assess the associations of potential causes of Alzheimer’s disease, i.e., systemic inflammatory regulators, plasma amyloid, cerebrospinal fluid (CSF) amyloid and tau, then potential causes of inflammation, i.e., testosterone, sex-hormone binding globulin (SHBG), insulin-like growth factor 1 (IGF-1), autoimmune diseases and finally more broadly the role of the plasma proteome in risk of Alzheimer’s disease. Genetic predictors of these risk factors and genetic associations with Alzheimer’s disease were obtained from published genome-wide association studies (GWAS). Genetic associations with Alzheimer’s disease were obtained from the GWAS of the International Genomics of Alzheimer's Project for clinical late-onset Alzheimer’s disease, the UK Biobank GWAS of family history of Alzheimer’s disease of parental proxy cases, as well as the UK Biobank summary statistics of siblings’ proxy cases. Possible selection bias from selective survival was also considered. No association with Alzheimer’s disease was found for the 41 systemic inflammatory regulators available, nor for plasma amyloid, CSF amyloid or tau, although liability to Alzheimer’s diseases was associated with higher basic fibroblast growth factor, granulocyte colony-stimulating factor, interferon gamma, interleukin-7, interleukin-13, interleukin-2 and tumor necrosis factor-alpha, and lower interleukin-18. Testosterone could be inversely associated with Alzheimer’s disease but not SHBG or IGF-1. Liability to some immune system conditions could be related to Alzheimer’s disease, i.e., a positive association of multiple sclerosis and an inverse association of liability Sjogren's Syndrome but not for psoriasis or sarcoidosis. Associations for rheumatoid arthritis, type 1 diabetes and giant cell arteritis were inconsistent. Of the 3282 plasma proteomes assessed, 29 proteins were associated with Alzheimer’s disease. Positive associations of cardiotrophin-1, leucine-rich repeat neuronal protein 1; inverse association for protein S100-A13 and tubulin-specific chaperone A were found with risk of siblings’ Alzheimer’s disease, while myeloid cell surface antigen CD33 was positively associated with clinical Alzheimer’s disease. With consideration of possible selection bias due to selective survival, this thesis found evidence for inverse associations of testosterone, liability to Sjogren's Syndrome, protein S100-A13 and tubulin-specific chaperone A; and positive association of liability to multiple sclerosis, cardiotrophin-1, leucine-rich repeat neuronal protein 1 and myeloid cell surface antigen CD33 with risk of Alzheimer’s disease. These results suggested a possible role of neuroinflammation and myelin damage which may be investigated in future studies.