Lipoprotein(a) and Cardiovascular Risk

Evidence Review

The case that Lp(a) is a causal — not merely associated — risk factor for cardiovascular disease has been built from three lines of evidence: large prospective epidemiology, Mendelian randomisation, and now early therapeutic trials.

The strongest epidemiological evidence comes from the Emerging Risk Factors Collaboration meta-analysis by Erqou and colleagues. Pooling individual records from 36 prospective studies including 126,634 participants and 1.3 million person-years of follow-up, the collaboration documented 22,076 vascular and non-vascular outcomes. After adjustment for age, sex, and conventional risk factors, the hazard ratio per one standard deviation higher log-Lp(a) was 1.13 (95% CI 1.09–1.18) for coronary heart disease and 1.10 (95% CI 1.02–1.18) for ischaemic stroke. The associations were continuous, modest at the level of individual standard deviations, but substantial at the upper extreme of the distribution where Lp(a) levels can be 10- to 100-fold above the population mean. Crucially, the relationship was specific: there was no association with non-vascular mortality, indicating Lp(a) was not just a marker of generalised ill health. [2]

The Mendelian randomisation evidence by Kamstrup and colleagues used the LPA gene's natural genetic variation as an unbiased instrument. The kringle IV type 2 (KIV-2) repeat polymorphism within LPA inversely determines Lp(a) concentration: fewer repeats produce smaller apo(a) isoforms, faster hepatic secretion, and higher plasma Lp(a). Because KIV-2 alleles are randomly assigned at conception (independent of lifestyle confounders), differences in Lp(a) by KIV-2 genotype reflect lifelong differences attributable to Lp(a) itself. In 16,156 Copenhagen General Population participants and 9,330 Copenhagen City Heart Study participants, observational analysis showed a multivariable-adjusted hazard ratio of 2.6 for myocardial infarction in the highest Lp(a) quintile, while genetic analysis using KIV-2 produced a corresponding causal odds ratio of 1.22 per doubling of Lp(a) — highly consistent with the observational data and unconfounded by lifestyle. [3] This study, replicated in multiple cohorts and meta-analysed across hundreds of thousands of participants, established that Lp(a) is a causal driver of myocardial infarction, not merely a passive marker.

The valvular evidence comes from two complementary studies. Thanassoulis and colleagues conducted a genome-wide association study in 6,942 participants from the CHARGE consortium for aortic-valve calcification by CT scan, with replication in 28,193 participants for incident clinical aortic stenosis. The single SNP rs10455872 in LPA — which raises Lp(a) about 4-fold — showed an odds ratio of 2.05 (per allele) for valve calcification across white, African-American, and Hispanic-American participants, with a hazard ratio of 1.68 for incident aortic stenosis in Swedish prospective follow-up, replicated independently in Danish data. [5] Kamstrup and colleagues then quantified the dose-response in 77,680 Copenhagen general-population participants: relative to Lp(a) <20 mg/dL, hazard ratios for incident aortic stenosis were 1.6 at 20–64 mg/dL, 2.0 at 65–90 mg/dL, and 3.0 at >90 mg/dL, with corresponding odds ratios from Mendelian-randomisation analyses showing genetic concordance. [6] These two studies established Lp(a) as the first identified causal driver of calcific aortic valve disease, transforming aortic stenosis from a "wear and tear" condition into a partly genetic, partly modifiable disease in development.

The niacin meta-analysis by Sahebkar and colleagues addressed the historical hope that B3 might be the answer. Pooling 14 randomised placebo-controlled trials with 9,013 participants (5,362 randomised to extended-release niacin), the meta-analysis demonstrated a weighted mean Lp(a) reduction of 22.9% (95% CI −27.3 to −18.5, p<0.001). [7] The Lp(a)-lowering effect is real and reproducible. However, the major outcomes trials AIM-HIGH (n=3,414) and HPS2-THRIVE (n=25,673) tested niacin added to statin therapy in patients with established cardiovascular disease and showed no reduction in cardiovascular events, while increasing rates of bleeding, infection, and new-onset diabetes. The disconnect between Lp(a) lowering and event reduction is most likely because niacin's effect is partial, statin background already optimised LDL, and niacin's other off-target effects offset the benefit. The clinical conclusion is that niacin should not be used to reduce cardiovascular risk via Lp(a) lowering.

The diet evidence by Faghihnia and colleagues is small but mechanistically important. In a tightly controlled crossover feeding trial of 63 healthy adults, four weeks on a low-fat (20% fat, 5% saturated) high-carbohydrate diet raised mean Lp(a) by approximately 12% relative to a high-fat (40% fat, 13% saturated) diet, while concurrently lowering LDL cholesterol and increasing oxidised-phospholipid burden on apoB lipoproteins. [8] The signal has been replicated in subsequent feeding trials including the GET-READI study in African Americans. The clinical implication is not to recommend high saturated-fat intake — outcomes data still favour Mediterranean and DASH-style patterns — but rather that strict low-fat dieting in someone with already-elevated Lp(a) may worsen one risk axis while improving another, and that "low fat" should not be the default lifestyle prescription for high-Lp(a) patients.

The landmark Tsimikas review in JACC summarises the integrated case for Lp(a) as an independent, genetic, likely causal risk factor for cardiovascular disease and calcific aortic valve stenosis, and articulates the rationale for testing once in every adult and considering Lp(a)-targeted therapy in high-risk patients. [1] The review preceded — and helped motivate — the antisense oligonucleotide proof-of-concept trial published by Tsimikas and colleagues in NEJM in 2020, in which 286 patients with established cardiovascular disease and screening Lp(a) ≥60 mg/dL were randomised to subcutaneous AKCEA-APO(a)-LRx (now pelacarsen) at multiple doses or placebo. The drug produced dose-dependent Lp(a) reductions of 35% (20 mg every 4 weeks) up to 80% (20 mg every week) versus 6% with placebo, with no significant safety signals on platelets, liver, or kidneys, and the most common adverse events were injection-site reactions. [4] The phase 3 HORIZON trial is now testing whether pelacarsen translates this Lp(a) reduction into reduced major adverse cardiovascular events — the field's first definitive test of "Lp(a) lowering = event reduction." Companion trials of small interfering RNA agents (olpasiran, lepodisiran, zerlasiran) are showing 80–98% Lp(a) reductions with quarterly or biannual dosing.

Overall confidence assessment: Lp(a) is among the most thoroughly established causal cardiovascular risk factors known, supported by epidemiology in over a million participants, multiple Mendelian-randomisation studies, and a coherent mechanistic biology spanning atherogenesis, thrombosis, and aortic valve calcification. The clinical recommendation to test once in adulthood is supported by every major lipid society. The evidence base for intervention is in transition: niacin is settled as ineffective, statins are not Lp(a)-targeted, PCSK9 inhibitors lower Lp(a) modestly as a beneficial side effect of LDL reduction, and the new RNA-based agents are highly effective at lowering Lp(a) but await outcome confirmation in 2026–2027. Until then, the evidence-based response to a high Lp(a) result is to know your number, drive every other modifiable risk factor — LDL, blood pressure, smoking, diet, fitness, weight — to optimal levels, and discuss with a cardiologist or lipidologist whether you are a candidate for emerging therapies as they become available.