USE OF MENDELIAN RANDOMIZATION STUDIES TO IDENTIFY POSSIBLE PHARMACOLOGICAL TARGETS IN THE CARDIOVASCULAR AREA

Different statistical approaches have been implemented to overcome the limitations that typically and differently influence both randomized clinical trials and observational studies. Mendelian randomization studies, in which functional genetic variants serve as tools (“instrumental variables”) to approximate modifiable environmental exposures, have been developed and implemented in the context of observational epidemiological studies to strengthen causal inferential estimates in non-experimental situations. Since genetic variants are randomly transferred from parents to offspring at the time of gamete formation, they can realistically mimic the random allocation process of treatment in a randomized clinical trial, offering a strategy to eliminate, or at least reduce, the residual confounding typically affecting observational studies, thus allowing to obtain generalizable results for the entire population. If correctly conducted and carefully interpreted, Mendelian randomization studies can provide useful scientific evidence to support or reject causal hypotheses that verify the association between modifiable exposures and diseases. This kind of evidence may provide useful information to identify new potential drug targets, with a higher probability of success than approaches based on animal models or in vitro studies. This thesis summarizes the history and context of Mendelian randomization, the main features of the study design, the assumptions for its correct use, and a brief discussion on the advantages and disadvantages of this approach. In addition, an overview of what the Mendelian randomization technique has contributed to date in the cardiovascular field has also been presented. The methods and techniques discussed have been also practically applied on several studies conducted thanks to a collaboration established with Professor Brian A. Ference from the Cardiovascular Epidemiology Unit of the Department of Public Health and Primary Care (University of Cambridge). This agreement has allowed to access the UK Biobank, a prospective cohort study with deep genetic, physical, and health data, collected on about 500,000 volunteer participants recruited throughout the UK. The access to this large-scale biomedical database has been fundamental to carried out the projects presented in this thesis, which have provided key evidence to improve our knowledge about cardiovascular disease. First, we found that the increase of measured body mass index is a much stronger risk factor for type 2 diabetes than polygenic predisposition that leads to reversible metabolic changes that do not accumulate over time. Therefore, most cases of diabetes potentially can be prevented or reversed, leading to a major reduction of the prevalence of one of the most impactful risk factors for the development of cardiovascular disease. Second, we found that parental family history of coronary heart disease provides independent, complementary and additive information to the individual polygenic predisposition in the definition of the inherited genetic variation as well as to LDL cholesterol exposure in the estimation of the lifetime cardiovascular risk. In order to develop a simple, but powerful, algorithm to contextualize the frame of who will need to be treated, it is essential to retrieve information about parental family history of heart disease and individual polygenic predisposition to coronary artery disease, in addition to the measurement of all the other well-known cardiovascular risk factors, especially LDL cholesterol levels. Finally, we discovered three important evidence regarding lipoprotein(a), an independent risk factor for the development of coronary and cerebral atherosclerosis: (i) the cumulative lifetime risk of major coronary events is comparable considering genetically and clinically determined Lp(a) concentrations, meaning that, in terms of cardiovascular risk prediction, it is reasonable to rely on measured levels, regardless the genotype; (ii) there is no significant association between high Lp(a) concentrations and the occurrence of venous thromboembolism events; (iii) an extra reduction of LDL cholesterol can overcome the extra cardiovascular risk due to high Lp(a) levels, and we quantitatively defined the additional LDL cholesterol reduction needed to abolish this risk. At the end of this dissertation, the potential use of Mendelian randomization to inform the design of randomized controlled trials is also presented, as well as the possibility to use this approach to anticipate trials results in terms of predicting treatment efficacy and adverse effects, and to inform on potential repurposing of drugs.