Estimation of required reduction of Lp(a) for a clinically relevant reduction of CHD risk
Estimation of the Required Lipoprotein(a)-Lowering Therapeutic Effect Size for Reduction in Coronary Heart Disease Outcomes - A Mendelian Randomization AnalysisLiterature - Lamina C, Kronenberg F, for the Lp(a)-GWAS-Consortium - JAMA Cardiol. 2019. doi:10.1001/jamacardio.2019.1041
Introduction and methods
Genetic studies have provided support for causality of the link between Lp(a) concentrations and higher coronary heart disease (CHD) risk [1-4]. Therapy to lower Lp(a) levels is in development, and antisense oligonucleotides that lower Lp(a) production by up to 90% are now available .
To set up interventional studies with Lp(a)-lowering drugs, it is important to estimate the required lowering of Lp(a) to efficiently improve clinical outcomes. Estimates have been published of 50-60 mg/dL  to more than 100 mg/dL  to produce a risk reduction similar to an LDL-c lowering of 38.67 mg/dL.
This study used a mendelian randomization approach to estimate the required lowering of Lp(a) that would show the same risk association with CHD risk lowering as a 38.67 mg/dL therapeutic lowering of LDL-c level. Data of a GWAS meta-analysis on Lp(a) published by the same authors in 2017 were used, based on 5 primarily population-based studies (n=13781). The Lp(a) concentrations were all measured in the same laboratory. Median values for all studies ranged from 11 to 12 mg/dL, except for Finnish patients, who are known to have half the levels seen of other white populations.
27 single-nucleotide polymorphisms (SNPs) were included. An OR for LDL-c (in 10 mg/dL) of 0.855 (95%CI: 0.818-0.893) was taken from a mendelian randomization analysis of LPA variants by Burgess et al. , which is used in this analysis to calculate an OR for Lp(a) via the formula: 38.67 x log (OR for LDL-c)/log (OR for Lp[a]). When assuming that the relation between short-term and life-long risk reduction is similar for LDL-c and Lp(a), an estimate for short-term trials can be deducted from the genetically predicted estimate.
- Based on the 27 included SNPs, an OR for CHD risk of 0.912 was estimated for each 10 mg/dL lower genetically predicted Lp(a). This corresponds to a decrease of Lp(a) by 65.7 mg/dL to be similar to a decrease in LDL-c by 38.67 mg/dL.
- Genetically estimated 10 mg/dL lower Lp(a) levels were associated with an 8.8% (95%CI: 7.5-10.1) lifetime lower risk of CHD, and with a short-term lower risk of 3.7%.
- The short-term risk reduction of 22% with LDL-c lowering is based on a median of 5 years of treatment in the CTT trial, and the data suggest 45% risk reduction over the lifetime based on genetically caused reduction of LDL-c.
- The data suggest that Lp(a) would have to be reduced by 65.7 mg/dL (95%CI: 46.3-88.3) for the risk reduction to be comparable with the CHD risk reductions associated with 38.67 mg/dL LDL-c reduction.
This mendelian randomization analysis suggests that lowering Lp(a) by 65.7 mg/dL with an Lp(a)-lowering therapy would be needed to reach the same potential effect on clinical outcomes as lowering LDL-c levels by 38.67 mg/dL with therapy.
It should be noted that this finding is based on the assumption that the lifetime risk reduction based on genetic effects compared with the short-term risk reduction based on therapeutic lowering of LDL-c shows the same ratio for LDL-c and Lp(a). It is, however, to date unknown whether the pathophysiological effects and mechanisms of LDL-c and Lp(a) in the development of atherosclerosis are similar.
Thanassoulis  agrees that, because of the recent development of Lp(a)-lowering drugs, the question of how much Lp(a) should be lowered, and for how long, has become even more relevant.
Lipoprotein(a) dyslipidemia [Lp(a) >50 mg/dL] is the most common genetic dyslipidemia worldwide, which is associated with a higher lifelong risk for MI, stroke and aortic stenosis. Lp(a) levels are largely determined by a person’s genetic make-up: many genetic variants exist that may raise an individual’s plasma Lp(a) level. Thus, the level is Lp(a) is hardly affected by diet, exercise and other lifestyle and environmental differences. Thus, on average, Thanassoulis emphasizes ‘there will be no differences, whether via other genes or environment, between people who inherit many Lp(a) variants who have high Lp(a) levels compared with those who inherit few and have low levels.’ This natural (or mendelian) randomization of Lp(a) variants can benefitted from to evaluate the causal role of Lp(a).
Thanassoulis concludes that, based on Lamina’s estimate of a necessary Lp(a) lowering of 65.7 mg/dL to provide a 22% short-term coronary risk reduction, and because novel Lp(a)-lowering agents have been shown to lower Lp(a) by up to 90%, the mean Lp(a) level of patients who enroll in future trials will need to be greater than 75 mg/dL. Assuming similar event rates as in statin trials, Lp(a) trials should last at least 5 years and include similar number of patients as in the secondary prevention statin trials.
Thanassoulis lists a few caveats of this approach, including that the estimates reflect uncertain approximations. If the true Lp(a) lowering that is required for a clinically meaningful effect is greater, this has consequences for the minimum level at enrollment. It has been argued before that, because of the number of assumptions involved in generating estimates, mendelian randomization is more suitable to provide evidence for or against causality than to estimate quantitative treatment effects. The estimates also assume that Lp(a) has a similar chronic cumulative effect as LDL-c. It is, however, suspected that Lp(a) has a direct role in thrombosis. If this is indeed the case, then the degree and duration of Lp(a) lowering needed for benefit would be less than the estimated amount. Moreover, the estimates were based on a cohort of people potentially experiencing a first incident event. Future initial RCTs are likely conducted in individuals with prevalent coronary artery disease. It remains to be established whether the degree of required Lp(a) lowering is greater in the latter population.
Despite these caveats, Thanassoulis concludes that ‘it is nonetheless remarkable that we can validate novel therapeutic targets such as Lp(a) and also resolve key trial parameters prior to starting any RCT, based on such genetic studies.’ Still, the ultimate test of the Lp(a) hypothesis will be a true RCT of Lp(a) lowering.