Possible Reasons Why the Lp(a)HORIZON Lipoprotein(a)-Lowering Trial Failed to Demonstrate Improved Clinical Outcomes: Implications for Laboratory Medicine

The Lp(a)HORIZON trial aimed to determine whether lipoprotein(a) [Lp(a)] reduction using the drug pelacarsen could reduce cardiovascular (CV) risk in patients with both established cardiovascular disease (CVD) and an Lp(a) level ≥70 mg/dL (approximately 149 nmol/L), whose other CV risk factors were otherwise already treated optimally (1). On September 4, 2026, the topline results of the trial were announced stating that the trial failed to meet its primary endpoint of reducing the risk of CV events (2). This finding raises questions around the future of Lp(a) lowering as a target for improving CV outcomes and the benefit of measuring Lp(a) in this patient group. Given that Lp(a) resembles both LDL and plasminogen, it is understandable that it has gained attention as a CV risk factor or marker, but this interest has varied over the decades. In 1993, the negative findings in the large Physicians’ Health Study (3) were influential and lessened enthusiasm for Lp(a) at a time when the benefits of LDL cholesterol (LDL-C) lowering with statin treatment were being so compellingly demonstrated. This mood persisted despite a sample reanalysis of the Physicians’ Health Study, which showed more positive findings after using an Lp(a) assay that was relatively unaffected by apolipoprotein(a) size polymorphisms (4), thereby supporting the notion that Lp(a) concentrations be reported in relation to particle numbers (the molar unit being nmol/L) rather than as mass (mg/dL). Subsequently, Lp(a) underwent a sustained renaissance when Mendelian randomization studies consistently showed that the largely genetically determined concentrations of Lp(a), which remain similar throughout life, were most likely a causal risk factor rather than simply a risk marker for CVD risk (5). Other epidemiological studies, in contrast to the Physicians’ Health Study, consistently showed a clear association between high Lp(a) concentration and CV risk (6). As a consequence, many authoritative national and international guidelines now recommend measuring Lp(a) at least once in an adult’s life, with a view to using it as an additional factor to assess CV risk and, if raised, to inform cascade screening of family members. More recently, subcutaneously injectable treatments (antisense oligonucleotides and small interfering RNA), which target the mRNA coding for apolipoprotein(a), thereby preventing assembly of Lp(a) and so substantially lowering blood concentrations, have been developed (7). Pelacarsen, an antisense oligonucleotide, is the first of these agents to report outcome data (2). This placebo-controlled double-blind study tested pelacarsen against placebo in a 1:1 ratio in 8323 patients with Lp(a) concentrations ≥70 mg/dL who had existing optimally treated CVD (1). The 2 primary outcomes were to assess whether pelacarsen could reduce major adverse CV events (MACE) in the full study group and/or a subgroup with a screening Lp(a) ≥ 90 mg/dL. Pelacarsen was administered monthly, intended to produce an 80% reduction in circulating Lp(a) during a follow-up period of >2.5 years. The study was powered to detect a 20% reduction in its primary endpoints with >90% power (1). Given the promise that Lp(a) is such a potent risk factor for CVD, why might the Lp(a)HORIZON trial have been unable to show any benefit? Existing lipid-modifying agents influence Lp(a) concentrations, although not to the extent of the more recent injectable treatments. Agents with the largest reductions are the cholesteryl ester transfer protein inhibitors, with a 46% reduction from baseline, niacin (nicotinic acid) with a 37% reduction, and proprotein convertase subtilisin/kexin type 9 inhibitors showing a 29% reduction (8). However, niacin was withdrawn from sale as a lipid-lowering agent because trial data showed no improvement in CV outcomes (9), and cholesteryl ester transfer protein inhibitors have not been pursued commercially, beyond their LDL-C–lowering properties, because of both a lack of efficacy and safety concerns with some agents (10). In contrast, Proprotein convertase subtilisin/kexin type 9 inhibitors have consistently reduced MACE, although a meta-analysis of one of these agents showed Lp(a) reduction not to be associated with MACE independently of LDL-C reductions (11). Paradoxically, the classes of drugs with the best evidence base for reducing CV risk, namely the statins, ezetimibe and bempedoic acid, do not alter Lp(a) concentrations to the same degree (8). This disparity between the relative ability of these drugs to lower Lp(a) and their effectiveness at reducing CV risk has seldom been highlighted but may, in retrospect, have been a cautionary observation and seems of particular relevance following the outcome of Lp(a)HORIZON. In the 21st century, other credible markers of CV risk have unfortunately not fulfilled their promise. Among these are HDL cholesterol and plasma homocysteine. Cholesteryl ester transfer protein inhibitors addressed the hypothesis that raising HDL cholesterol by 30% to 133% could complement reducing LDL-C in lowering CV risk, but as previously mentioned, had minimal influence on CV outcomes (10). Enthusiasm for plasma homocysteine in the early 2000s, a test which also had strong epidemiological evidence extending to causality, initially followed a similar trajectory to that of Lp(a) in recent years (552 PubMed articles with “homocysteine” in their title in 2005 and 550 with Lp(a) or spelling variants in 2025). Intervention to reduce CV risk, this time with B vitamins, again proved unsuccessful (12). Treatments for hypertension, lipids and hyperglycemia can have an influence on their respective measured markers (blood pressure, LDL-C and glucose) within days or weeks of starting medication, but their subsequent impact on reducing CV risk is over different timescales, being within months to a year for antihypertensives, within a year to multiple years for lipid lowering, and multiple years to decades following improvements in glycemia. It is therefore speculation to suggest that lowering Lp(a), after a lifetime of exposure, would necessarily lead to any reduction in CV risk over a typical 5-year clinical trial or ever. Initial trials targeting LDL-C reduction often comprised patients with existing coronary or cerebrovascular disease (secondary prevention), as their high frequency of recurrent events allowed treatment effectiveness to be established with fewer patients over a shorter period. This population was also chosen for the Lp(a)HORIZON study, having a prior history of myocardial infarction, ischemic stroke, or symptomatic peripheral arterial disease (1). This apparently sensible choice may have inadvertently reduced the chance of demonstrating a treatment effect since the epidemiological CV risk gradient of Lp(a) seems steeper in subjects without CVD [hazard ratio, 1.10 for every 50 nmol/L (approximately 25 mg/dL) increase in Lp(a)] than in those with existing CVD (hazard ratio, 1.05) (13). This slope difference could demonstrate that optimal control of other CV risk factors leads to an attenuation in the residual risk explained by Lp(a). Added to this, epidemiological studies suggest Lp(a) is more closely associated with recurrent coronary artery disease than with ischemic stroke, and either event is more common in those with baseline coronary artery disease rather than previous stroke (13). In the Lp(a)HORIZON study, 9.8% of participants were recruited because they had a previous ischemic stroke, and the definition of MACE included fatal and nonfatal stroke (1). Each of these stroke patient inclusions may therefore have led to an apparent reduction in treatment effect. In contrast, other Lp(a)-lowering trials using different agents, such as olpasiran in the OCEAN(a) study, only include patients with previous coronary disease and have primary outcome measures influenced solely by coronary artery disease events (7). The fact that Lp(a)HORIZON recruited using a mass rather than molar Lp(a) assay could have unintentionally led to a cohort less likely to show treatment benefit. However, while these assays can categorize individuals quite differently, a recent reanalysis of Lp(a)HORIZON samples showed 97% agreement when categorizing patients using the 70 mg/dL trial threshold compared with using a 149 nmol/L molar cutoff (14). It suggests that assay units may not be a large factor in the trial failure. Had the Lp(a)HORIZON study shown a compelling beneficial effect of Lp(a) lowering on reducing CV risk, then it would likely have heralded a rapid expansion of laboratory requests [from its current test prevalence of approximately 0.1% in a US population (15)] to include most individuals with established CVD and for monitoring Lp(a) treatment response. There would also have been a further incentive to screen family members for high values. However, a single negative trial announcement does not mean increased demand will never materialize. For instance, it is not yet known if there was any signal of benefit to Lp(a) lowering in the trial just short of statistical significance, rather than there being completely no sign of CVD reduction. There are also 3 other Lp(a) secondary prevention intervention studies scheduled to report in the next 5 years (7), 2 of which use treatments that can lower Lp(a) concentrations more than pelacarsen. With different study designs and agents that can reduce Lp(a) further, they may be more likely to show a treatment benefit. It is also possible that it is the primary prevention population (without existing CVD) who might benefit most from Lp(a) lowering given their steeper CV risk gradient, but outcome data for this group are unlikely until at least 2031. At the moment, it seems prudent to continue using Lp(a) testing as currently recommended. If subsequent studies do not show CV risk to be modifiable by lowering Lp(a), then it will be interesting to see whether the test continues to be used to help guide the decision to start or intensify more established CV treatments such as statins and antihypertensives. However, based on the frequency of requests for other CV risk markers with similar potential purposes, such as high-sensitivity C-reactive protein or plasma homocysteine, even this continued use might not be guaranteed. Nonstandard Abbreviations: Lp(a), lipoprotein(a); CV, cardiovascular; CVD, cardiovascular disease; LDL-C, low-density lipoprotein cholesterol; MACE, major adverse cardiovascular events. Author Contributions: The corresponding author takes full responsibility that all authors on this publication have met the following required criteria of eligibility for authorship: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; (c) final approval of the published article; and (d) agreement to be accountable for all aspects of the article thus ensuring that questions related to the accuracy or integrity of any part of the article are appropriately investigated and resolved. Nobody who qualifies for authorship has been omitted from the list. E.S. Kilpatrick (Conceptualization, Formal analysis [equal]) Authors’ Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the author disclosure form. Research Funding: None declared. Disclosures: E.S. Kilpatrick is a deputy editor for Clinical Chemistry (ADLM).