Lipoprotein(a) in various conditions: To keep a sense of proportions

Abstract

About two decades ago, a first large wave of epidemiological studies on lipoprotein(a) [Lp(a)] sloshed around the scientific community. Some of these data were so bad that the credibility of Lp(a) was almost drowned by them. Several reasons were responsible for this situation of contrasting results. First, at that time statistical packages became available which made it no longer necessary to be educated instatistics. A fewclicks separatedone fromthepublication, and data were often not analyzed to investigate a hypothesis but to create a hypothesis. Second, the pronounced skew of the Lp(a) distribution to the right, at least in non-African-American populations, with themajority of the individuals having low Lp(a) concentrations, requires a thorough handling of the data analysis with appropriate statistical procedures. Third, assays for measurement of Lp(a) were expensive and some of themwere of poor quality (particularly poor sensitivity), acting like a random generator of numbers unrelated to biological laws andphysiological regulatorycircuits. Andfinally, Lp(a) concentrationshave amore than1000-fold interindividual range that is to a large extent genetically determined [1]: up to 70% of the concentrations are explained by a highly polymorphic copy number variationwithin the LPA gene region thatwas alreadydescribedmore than 25 years ago [2,3]. These, and some other circumstances combined with hasty reactions targeting the “this is the first study” bonus, resulted often in a triplication of results and publications that described a negative, a positive or no correlation between Lp(a) and a certain factor x with a very flat learning curve. There were cross-sectional studies at that time comparing sometimes less than 20 patients with treatment A to patients with treatment B, which concluded, from a major