This may be the most shocking result in cholesterol research since the HDL drugs failed. A trial of over 8,000 people tested a simple idea. If we lower a particle in our blood called Lp(a), which is thought to be especially harmful to our blood vessels [1], we should see less heart disease. But it didn't. It failed [2].
So have we got the cholesterol hypothesis wrong? For decades we've been taught that lowering the cholesterol in our blood reduces heart disease. So what's going on? And if your own Lp(a) is high, what should you do now?
Table of Contents
- What Lp(a) is
- The Lp(a)HORIZON trial
- Could Lp(a) be like HDL?
- Five reasons the trial may have failed
- The aortic valve
- My take
- Have we got the cholesterol hypothesis wrong?
- References
What Lp(a) is
Every cell in your body makes its own cholesterol [3]. But sometimes cholesterol needs to be moved around the body, to tissues that need a temporary top-up. And cholesterol is fatty. It doesn't mix with water. So it gets packaged up in a carrier called a lipoprotein [4].
Some of these lipoproteins are large, called remnants. Some are LDL particles, which you've heard of. And others are Lp(a). Each of these carries one copy of a protein called ApoB [5]. Even the large ones are small enough to get into the walls of our blood vessels, where they can become trapped and, over time, build up into blockages [6].
From mountains of genetic and observational data, we can see that higher Lp(a) goes with more heart disease. Your Lp(a) level is more than 90 percent set by your genes, and people born with high Lp(a) carry it for life [7]. So we can compare them with people born with low Lp(a), and see who gets more heart disease. About one in five people have high Lp(a). And yes, those people do have more heart disease [7, 5].
In nanomoles per litre, under 75 is normal, 75 to 125 is a grey zone, and above 125 is high [7]. At 125, the risk is about 40 percent higher. At 250, it's about double [5].
- European Atherosclerosis Society: normal is under 75 nmol/L (under 30 mg/dL), the grey zone is 75 to 125 nmol/L (30 to 50 mg/dL), and high is over 125 nmol/L (over 50 mg/dL) [7].
- 2026 American guideline: about 40% higher risk at 125 nmol/L (50 mg/dL), and about double at 250 nmol/L or more (100 mg/dL or more) [5].
The Lp(a)HORIZON trial
Here's what the Lp(a)HORIZON trial did. It enrolled 8,323 people at 797 sites in 42 countries. Every one already had disease of the heart or arteries, and an Lp(a) of roughly 150 nanomoles per litre or more [8, 9].
Half got a monthly 80 milligram injection of pelacarsen [8]. It sharply cuts how much the liver makes of a protein that Lp(a) is built from. In an earlier trial, doses like this cut Lp(a) by roughly 70 to 80 percent [10]. The other half got a placebo.
On the 4th of September, Novartis announced that the trial had missed its main goal. Its chief medical officer said the results "did not demonstrate" lower cardiovascular risk across the trial as a whole [2].
The full results are due at the American Heart Association meeting in Chicago on the 7th of November [11].
So why did it fail?
Could Lp(a) be like HDL?
We've been surprised like this before, with HDL. Since the Framingham study in the 1970s, we've known that people with high HDL cholesterol tend to have less heart disease [12]. Across more than 300,000 people, those in the top third for HDL had well under half the heart disease rate of those in the bottom third [13].
So drug companies built drugs to raise HDL, and between 2006 and 2017 they were tested in big trials. Raising HDL on purpose didn't lower heart disease [14, 15, 16]. One drug, torcetrapib, actually increased deaths [17], and niacin caused more serious side effects [18]. The one drug that helped a little, anacetrapib, also lowered the cholesterol carried on ApoB particles [19].
So now we say HDL is more of a marker of heart health. Changing the marker isn't the goal.
Could it be the same with Lp(a)?
Five reasons the trial may have failed
Here are 5 potential explanations for the failure, then we need to talk about aortic stenosis, which is a narrowing of the aortic valve that's associated with Lp(a).
First, my read is that the patients in Lp(a)HORIZON were already so well treated that most of them likely had low ApoB [9]. Once ApoB is that low, I suspect there isn't much left for an Lp(a) drug to prevent.
Second, the main thing the trial measured was a combination of cardiovascular death, heart attack, stroke and urgent procedures to reopen a heart artery. The trial was built to find about a 20 percent drop in those events [9, 20].
But as Marios Georgakis, a researcher in Munich who studies the genetics of heart disease, points out, that 20 percent target was partly based on genetic studies, which converted a drop in Lp(a) into an equivalent drop in LDL [21, 22].
But LDL tends to climb as we age, while Lp(a) stays largely stable for life. The European consensus says Lp(a) reaches adult levels by about age five [7]. If artery damage builds up from years of exposure, then lowering Lp(a) for a few years late in life removes a smaller share of your lifetime exposure than lowering LDL does [21].
If he's right, a 20 percent drop was too ambitious, and the real benefit may have been too small for a trial this size to detect.
Third, in two big trials of PCSK9 inhibitors, FOURIER and VESALIUS-CV, Lp(a) tracked with heart attacks but not with strokes [23, 24]. This trial counted strokes, which could dilute any benefit.
Fourth, pelacarsen may not have cut Lp(a) levels deep enough. Newer drugs in testing have cut Lp(a) by more than 90 percent in early trials [25, 26].
Fifth, the Lp(a) hypothesis is wrong.
But I don't think that's the correct conclusion yet for the reasons I've outlined. This trial has issues, and the genetic case is still strong.
Taking a look again at HDL, people born with genes that raise only their HDL don't have fewer heart attacks [27], but people born with genes for high Lp(a) do have more [28, 29].
So unlike HDL, Lp(a) still looks like a cause of heart disease [7], and lowering ApoB is still proven to prevent heart attacks [30, 31, 32].
But even if it looks like we won't get much additional benefit from lowering Lp(a) in terms of heart attacks and narrowing of our arteries, we might see benefits for the aortic valve.
The aortic valve
The aortic valve is the main valve that lets blood out of the heart. In aortic stenosis, the valve slowly stiffens and narrows, so the heart has to push harder, and in time it may need replacing. Lp(a) is very likely one of the causes. People born with genes for high Lp(a) get aortic stenosis more often [7, 33].
Lowering LDL hasn't helped once the valve has started to narrow. Three trials of statin-based treatment didn't slow it down [34, 35, 36]. Those trials all started after the valve had narrowed. A 2014 genetic study suggests LDL matters earlier, when calcium is building up in the valve and aortic stenosis first appears [37], while higher Lp(a) is also linked to a narrowed valve getting worse faster [38]. So even if lowering Lp(a) doesn't cut heart attacks much once ApoB is low, it might still help the valve. A trial testing an Lp(a) drug in people with aortic stenosis is running now. It won't report until around 2030 [39].
There's a different group of drugs called PCSK9 inhibitors that work by lowering ApoB. They help the liver pull LDL particles out of the blood [40, 41].
But they also lower Lp(a) levels by about a quarter [5]. They aren't approved for that, but it's a nice additional potential benefit of PCSK9 inhibitors.
And there's already a hint that they might help the valve.
When researchers looked back through the FOURIER trial, people on the drug had about half as many valve problems after the first year, though that's from just 63 cases [42]. These drugs lower both ApoB and Lp(a), and we don't know yet which one mattered.
My take
So here's my take on Lp(a).
Still measure your Lp(a) once [5].
If you have a high Lp(a), you still want to reduce your cardiovascular risk factors as much as possible [5]. Healthy weight, insulin sensitive, good diet and exercise, systolic blood pressure below 120 if you're not frail [43, 44]. And in clinic, I aim for ApoB below 50 mg/dL (0.5 g/L). That's where I go further than the guidelines [5].
And on that point, for most patients we start with a low-dose statin and add ezetimibe if needed. But for someone with high Lp(a), I'm quicker to add a PCSK9 inhibitor, though it's not licensed to lower Lp(a) [5].
And the main reason I think why you might want to lower Lp(a) is there might still be a small decrease in heart attacks from lowering Lp(a) that the failed HORIZON study wasn't powered to detect or the drug didn't lower it enough, and there might be a reduction in aortic stenosis.
And since July, in the US, there's been a PCSK9 inhibitor you swallow. It lowers LDL about as much as the injections do, and Lp(a) too [45, 46]. At list price, it costs about a third to a half less than the injections [47, 48, 49]. But it hasn't yet been shown to prevent heart attacks [50]. Its heart-attack trial is due to finish at the end of 2029 [51].
Given that PCSK9 inhibitors already reduce Lp(a) levels (again not licensed for this officially), I'm not convinced that we need a dedicated Lp(a) drug to reduce heart attack rates.
The exception might be people with very high Lp(a), around 430 nanomoles per litre, where the risk is about four times normal [5].
Where I can see a role is to try to prevent or slow aortic stenosis.
Have we got the cholesterol hypothesis wrong?
So, have we got the cholesterol hypothesis wrong? No. And if your Lp(a) is high, you don't need to wait for a new drug. Get your ApoB low with the medicines we already have, and start sooner rather than later.
References
1. https://pubmed.ncbi.nlm.nih.gov/38233012/
3. https://pmc.ncbi.nlm.nih.gov/articles/PMC9344793/
4. https://www.ncbi.nlm.nih.gov/books/NBK305896/
5. https://www.ahajournals.org/doi/10.1161/CIR.0000000000001423
6. https://pmc.ncbi.nlm.nih.gov/articles/PMC7308544/
7. https://pmc.ncbi.nlm.nih.gov/articles/PMC9639807/
8. https://pubmed.ncbi.nlm.nih.gov/40185318/
9. https://www.sciencedirect.com/science/article/pii/S0002870325001012
10. https://www.nejm.org/doi/full/10.1056/NEJMoa1905239
11. https://www.radcliffecardiology.com/news/aha-scientific-sessions-2026-late-breaking-trials-revealed
12. https://pubmed.ncbi.nlm.nih.gov/193398/
13. https://pubmed.ncbi.nlm.nih.gov/19903920/
14. https://pubmed.ncbi.nlm.nih.gov/22085343/
15. https://pubmed.ncbi.nlm.nih.gov/23126252/
16. https://pubmed.ncbi.nlm.nih.gov/28514624/
17. https://pubmed.ncbi.nlm.nih.gov/17984165/
18. https://pubmed.ncbi.nlm.nih.gov/25014686/
19. https://pubmed.ncbi.nlm.nih.gov/28847206/
20. https://clinicaltrials.gov/study/NCT04023552
21. https://x.com/MariosGeorgakis/status/2106025409372618921
22. https://pmc.ncbi.nlm.nih.gov/articles/PMC6481553/
23. https://www.ahajournals.org/doi/10.1161/CIRCULATIONAHA.118.037184
24. https://www.ahajournals.org/doi/10.1161/CIRCULATIONAHA.126.080999
25. https://pubmed.ncbi.nlm.nih.gov/36342163/
26. https://pubmed.ncbi.nlm.nih.gov/40162643/
27. https://pubmed.ncbi.nlm.nih.gov/22607825/
28. https://pubmed.ncbi.nlm.nih.gov/19509380/
29. https://pubmed.ncbi.nlm.nih.gov/20032323/
30. https://pubmed.ncbi.nlm.nih.gov/28444290/
31. https://pmc.ncbi.nlm.nih.gov/articles/PMC4187506/
32. https://pmc.ncbi.nlm.nih.gov/articles/PMC6439767/
33. https://pubmed.ncbi.nlm.nih.gov/24161338/
34. https://pubmed.ncbi.nlm.nih.gov/15944423/
35. https://pubmed.ncbi.nlm.nih.gov/18765433/
36. https://doi.org/10.1161/CIRCULATIONAHA.109.900027
37. https://pmc.ncbi.nlm.nih.gov/articles/PMC4280258/
38. https://pubmed.ncbi.nlm.nih.gov/26361154/
39. https://clinicaltrials.gov/study/NCT05646381
40. https://pubmed.ncbi.nlm.nih.gov/19020338/
41. https://pubmed.ncbi.nlm.nih.gov/12730697/
42. https://pubmed.ncbi.nlm.nih.gov/32347887/
43. https://pubmed.ncbi.nlm.nih.gov/26551272/
44. https://pubmed.ncbi.nlm.nih.gov/38945140/
45. https://pubmed.ncbi.nlm.nih.gov/41879224/
46. https://www.hcplive.com/view/enlicitide-fda-approval-ldl-c-reduction-coralreef-outcomes-trial
48. Amgen press release, 6 October 2025: evolocumab direct-to-patient monthly price
49. Regeneron and Sanofi press release, 11 February 2019: alirocumab reduced US list price
50. https://www.tctmd.com/news/fda-approves-enlicitide-oral-pcsk9-inhibitor-ldl-lowering
