ABSTRACT Lipoprotein(a) [Lp(a)], a highly atherogenic, prothrombotic, and proinflammatory lipoprotein, is a causal, independent, and highly heritable (h2=70-90%) atherosclerotic cardiovascular disease (ASCVD) risk factor that is elevated in 1.5 billion people globally. Despite significant stakes and emerging therapies that can reduce Lp(a) by >80%, few studies have comprehensively identified and characterized the potentially broad phenotypic effects of Lp(a). This major research gap overlooks opportunities to anticipate adverse effects of therapeutic Lp(a) lowering, illuminate mechanisms of action, and identify novel treatment indications. A second major research gap is the Eurocentric evidence base of Lp(a), which has persisted despite Lp(a) being recognized as one of the most variable ASCVD risk factors across populations. This research gap constrains the generalizability, relevance, and reach of evidence that informs Lp(a) clinical, regulatory, and public health decision making. A common factor underlying both research gaps is the absence of validated Lp(a) measures in broad studies with dense phenotypic data. We propose to address this obstacle by assembling a large and broad consortium with validated Lp(a) measures and genotypic data. To expand our consortium, we will measure Lp(a) in African, Polynesian, and South American cohorts using validated assays. Next, to facilitate causal inference in studies without measured Lp(a) but with dense phenotypic data, we will leverage our consortium and statistical genetics advances to construct highly accurate Lp(a) polygenic risk scores (PRS) in all populations. These PRS will then be projected into biobanks with dense genotypic and phenotypic data, but no Lp(a) measures. Finally, we propose a suite of causal inference studies that substitute Lp(a) PRS for measured Lp(a) and examine broad phenotypes. These studies enable well-powered (n=1,137,708), comprehensive, and generalizable causal inference investigations that are