Using genetic data to determine the effect of routinely measured blood cell traits on disease

Abstract

Blood cell traits (BCTs), including white blood cells (WBCs) and platelets, are commonly measured in a routine blood test or hospital visit. This is because there is a well-established relationship between BCTs and diseases that lead to death and disability. Most studies on diseases associated with BCTs have been observational, and therefore generally prone to confounding and reverse causation. Given the health burden of diseases observationally linked to BCTs, it is desirable to determine whether these relationships are causal. Mendelian randomization (MR) is a method in genetic epidemiology which uses people’s genetic data to provide a causal estimate between an exposure and an outcome. Therefore, the overarching aim of my thesis was to use MR to advance the knowledge on diseases associated with BCTs. To investigate this, I focused on three diseases, each having their own methodological challenges: Chapter 3 – colorectal cancer (CRC); Chapter 4 & Chapter 5 – P. falciparum malaria; Chapter 6 – deep vein thrombosis (DVT). In Chapter 3 I provided evidence that a higher eosinophil and lymphocyte count reduced the risk of CRC, and a follow-up MR analysis revealed a possible protective role for allergic disease in CRC development. In Chapter 4 I identified a subset of UK Biobank participants that correspond to the African continental ancestry group, allowing me to conduct a genome-wide association study of neutrophil count to P. falciparum malaria in Chapter 5. Here, the MR analysis showed limited evidence for a causal relationship between neutrophil count and severe malaria. Finally, in Chapter 6 I conducted a phenome wide MR study to identify novel risk factors for DVT, and a follow-up analysis identified that a protein predominantly present in platelets, plasminogen activation inhibitor 1 (PAI-1), mediates the relationship between adiposity and DVT risk.