Mendelian Randomization: Principles and its usage in Lp(a) research.

Epidemiological studies investigating the association between a biomarker and a disease have many limitations. The most prominent among these is that we cannot impute causality purely from a statistical association. If we observe an association, the biomarker might really be causal for the development of the disease, the association might be caused by a confounding variable or by reverse causation. With Mendelian Randomization (MR) methods, we have a potent tool at hand to derive evidence for a direct causal relationship. One of the core assumptions of MR studies is that genetic variants can be identified, which are strongly associated with the biomarker of interest, and can serve as an instrument indicating lifetime exposure. Since Lp(a) is primarily genetically determined by KIV-2 repeats, that in turn determine apo(a) isoform size, and by numerous single nucleotide polymorphisms (SNPs) and SNP-scores, this assumption is definitely fulfilled and it is probably one of the best phenotypes to be studied with Mendelian Randomization methods. The first studies evaluating the causal role of Lp(a) for cardiovascular diseases were performed in the early 1990s and more recently gained interest after several Lp(a)-increasing SNPs were identified in genome wide association studies. In this review, the principles behind MR methods are explained, together with their important role for Lp(a) research, particularly reconsidered in their historic context. MR methods have also been used to estimate the extent of Lp(a) reduction that would be required to yield a clinically meaningful reduction in outcomes in clinical intervention trials.