Human metabolic pathways as predictors for hypertension based on Estonian Biobank data

Abstract

The aim of this master’s thesis is to identify relevant human metabolic pathways for incident hypertension prediction in the Estonian Biobank. Hypertension is the most important preventable risk factor for cardiovascular disease, as well as mortality and all-cause morbidity. Studying the relationship between hypertension and metabolites can reveal important pathways and biomarkers for early diagnosis. Metabolites from an extensive mass spectrometry metabolomics dataset are mapped to human metabolic pathways. Principal component analysis is performed on each pathway and the first three principal components are chosen. Highly correlated components are removed as a result of the correlation analysis. Relevant pathway components for the prediction of incident hypertension are found using Cox proportional hazards models on right censored and left truncated data with age as time scale. As a result, it was found that both BMI and carbon metabolism pathway can help predict incident hypertension before the diagnosis. The results were validated against the previous research.