In a press release on 4^th September 2026 Novartis announced that the Lp(a) HORIZON Phase III trial for pelecarsen failed to meet its primary endpoint of reducing the risk of cardiovascular disease (CVD) in patients with elevated Lp(a)^[1]. Details of the trial design have been previously published^[2]. This was an unexpected result for clinicians and scientists in Lp(a) research, since the scientific literature provides compelling evidence that high Lp(a) concentrations are causally linked to CVD.

The first comments from experts on the HORIZON outcome suggest waiting for the full publication and subgroup analyses, as well as other ongoing clinical trials, before drawing conclusions about event reduction by lowering Lp(a). Here, we attempt to outline possible future directions for Lp(a).

Background

Most case-control and prospective studies conducted so far confirmed Lp(a) as a risk factor for CVD. One of the first studies published in 1981 measured Lp(a) by rocket electrophoresis in 76 male post-MI patients matched with 107 male controls^[3]. At a cut-off level of 30 mg/dl of Lp(a), an odds ratio of 1.75 for MI was observed, while, at a cut-off level of 50 mg/dl of Lp(a), an odds ratio of 2.3 was found. The highest rate of MI was found in patients with additional high LDL. After dividing patients according to Fredrickson Types (a standard procedure at that time), it was determined that all MI patients with Type-IIa (FH patients) had significantly elevated Lp(a) values compared to controls without MI. Importantly, MI patients with Type-IV hyperlipoproteinemia, a group that today would be classified as metabolic syndrome and/or T2DM, had lower Lp(a) as compared to controls without MI.

A large number of trials published in the following years further supported Lp(a) as a risk factor for CVD. In the Copenhagen City Heart Study published in 2008, 9,330 individuals were followed for 10 years and approximately 500 had an MI. Hazard ratio point estimates ranged between 1.1-3.7^[4]. The authors concluded that the MI risk increases exponentially with increasing plasma Lp(a) concentrations.

Genetic studies confirmed the atherogenicity of Lp(a). In one of these studies, Lp(a) was measured in 1,486 young men at the age of 18 years and a family history of MI was considered. An approximately 2.5-fold higher MI incidence was found in parents whose sons had Lp(a) values exceeding 25 mg/dl in contrast to parents of sons with less than 25 mg/dl (32% versus 13.4%)^[5]. Considering the high heritability of Lp(a), we considered this as a strong argument for the causality of Lp(a) as a risk factor for MI.

Mendelian Randomization studies published in the following years confirmed Lp(a) as an independent causal risk factor for CVD^[6]. Mendelian Randomization data were also used as a tool to calculate the clinical benefit of reducing Lp(a). In 2018, Burgess et al.^[7] published Lp(a) data from 62,240 CHD patients and 127,299 control individuals and concluded that “The clinical benefit of lowering Lp(a) is likely to be proportional to the absolute reduction in Lp(a) concentration. Large absolute reductions in Lp(a) of approximately 100 mg/dL may be required to produce a clinically meaningful reduction in the risk of CVD”. This is an important consideration for future trials.

In contrast to the above studies, some have shown that Lp(a) is not a significant risk factor for CVD. For example, the Physicians’ Health Study by Paul Ridker et al.^[8] followed close to 15,000 male physicians for 5 years, of whom 296 subsequently suffered an MI. The authors concluded that “These data do not support the use of Lp(a) level as a screening tool to define cardiovascular risk among this population”.

Questions remaining to be answered about Lp(a) include its physiological function, the site and mode of Lp(a) catabolism, the molecular mechanism of statin and PCSK9 effects on Lp(a) levels, and the significance of Lp(a) for Diabetes Mellitus.

Oxidative stress, e.g. during ATP production (oxidative phosphorylation), leads to peroxidation of phospholipids in cell membranes and lipoproteins. These oxidatively modified phospholipids (OxPL) need to be eliminated to avoid cellular damage. Lp(a) is one of the strongest plasma components that binds OxPL, which contributes to its atherogenicity^[9]. OxPL specifically bind to KIV10 of apo(a) where His-33 in this domain appears to be essential^[10]. If Lp(a) is very low or absent in plasma, OxPL also bind to apoB in LDL and VLDL. These OxPL in remaining apoB-LDL have been shown to be predictive for major adverse cardiovascular events (MACE)^[11].

There are other plasma proteins that bind negatively charged phospholipids, whether oxidized or not, such as beta-2-glycoprotein-I^[12]. At high concentrations, Lp(a) is trapped in arteries and its OxPL attract inflammatory cytokines and promote plaque formation. The question remains whether reducing Lp(a) to very low levels in combination with significant LDL-reduction creates a situation where OxPL might not be removed effectively enough from plasma by other pathways. Measuring OxPL should certainly be considered in future Lp(a) trials.

It is also noteworthy that Lp(a) might have some beneficial effects, as it selectively binds PAF-AH (platelet-activating factor acetylhydrolase), the enzyme that inactivates PAF, one of the strongest triggers of platelet aggregation^[13].

Possible Reasons Why the HORIZON Trial Did Not Meet Its Primary Endpoints

• It is important to point out that certain components of the primary endpoint may have shown more reduction than others, so those not showing reduction in risk could have contributed to the primary endpoint not being significant. Also, it will be a different discussion if the differences are completely null versus borderline.

• Longer treatment periods and/or larger Lp(a) reductions may be required to show MACE reductions.

• In HORIZON, Lp(a) cut-offs were lower than in other clinical trials.

• Once other CV risk factors, especially LDL and apoB, are reduced to very low levels, the observed reductions in Lp(a) may not result in significant changes in the events.

• In a secondary prevention population such as HORIZON, platelet therapy with aspirin, clopidogrel, ticagrelor and prasugrel may reduce the atherothrombotic risk of Lp(a) also in the control group.

• Lp(a) in the HORIZON trial was analysed with the Tina-quant Lp(a) Gen 2 turbidimetric assay that individually measures Lp(a) either in mg/dl or nmol/L. The assays use polyclonal antibodies against apo(a) that by nature cause artifacts due to genetic polymorphisms in KIV2. In order to account for the size polymorphism in apo(a), Lp(a) values are calculated by an algorithm designed by the manufacturer. We know that such an algorithm is just an approximation and does not account for all polymorphic forms of apo(a). This may have affected the pelecarsen and the control group disproportionally.

• Patient selection and treatment duration may also be important, as patients with genetic dyslipidemias such as FH or Type-IV hyperlipoproteinemia may respond differently to Lp(a) lowering. Also, patients with MI, stroke and PVD may respond differently to Lp(a)-lowering strategies.

• Apo(a) exists in a great number of polymorphic forms, and these mutations may not have been fully considered in HORIZON. There is evidence that some of these polymorphisms exhibit significant differences in their atherogenicity.

• Such polymorphisms may also affect OxPL binding, binding to proteoglycans and clotting factors, catabolic pathways and more.

• The negative findings of the trial may in part be due to the selection criteria of the recruited patients.

To Fully Judge about the Significance of the Lp(a) HORIZON Outcome the Following Additional Information is Required

• Differences in OxPL between patients treated with pelacarsen and controls.

• Subgroup analyses of patients on lipid-lowering drugs that may have an effect on Lp(a) concentrations (PCSK9 inhibitors, niacin, etc.).

• Subgroup analysis in individuals with T2DM, metabolic syndrome and high triglycerides.

• Subgroup analyses in patients with very high starting LDL-C (FH-patients).

• LDL-C values corrected for Lp(a) cholesterol in pelacarsen and control patients.

Future Implications

It will be important to advise health care providers and patients concerning the outcome of the Lp(a) HORIZON trial and to emphasize that the information provided by Novartis so far is fragmentary. It will be important to await the results of the subgroup analysis as well as results of additional Phase III trials before final conclusions are drawn. In the meantime, the following points should be considered:

• Measuring Lp(a) needs to be continued as this allows CV risk to be assessed more accurately. A rough estimate might also be made of how much of the measured “LDL-C” concentration is bona fide LDL-C and what portion might be the contribution of Lp(a)-C.

• Phase III clinical trials that are currently underway are performed with medications that are more efficient on one hand or are based on completely different molecular mechanisms. It is hoped that these trials will give positive results.

• Although the HORIZON trial did not meet its primary endpoint, Lp(a) remains an important, independent risk factor for CVD. Apheresis trials and subgroup analyses from PCSK9 inhibitor studies indicate that lowering Lp(a) reduces CVD risk.

• Patients with high Lp(a) levels need special attention, and LDL-C and other established risk factors should be treated more aggressively.

Authors contribution

Kostner G, Kostner K: Writing-original draft, writing-review & editing.

Conflicts of interest

Gerhard Kostner and Karam Kostner are the Editorial Board members of Advances in Lipoprotein(a) Research.

Ethical approval

Not applicable.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Availability of data and materials

Not applicable.

Funding

None.

Copyright

© The Author(s) 2026.