Why doesn't Lp(a) go down with diet?
Lp(a) doesn't go down with diet because its level depends on how much the liver makes, which is set more than 90% by a single gene, LPA, and fixed by around age 5. Diets act mainly on clearing LDL: eating less saturated fat even raises Lp(a) by 10 to 15%.
Why does the Lp(a) level depend on a single gene?
Simplified model to show the direction of the effect: the levels in the drawing do not predict any one person's level.
The model in text, and where these numbers come from
| Pattern copies on one gene | Chains leaving the liver (model) | Lp(a) supplied by this gene (model) |
|---|---|---|
| 3 | 100 % | about 160 nmol/L |
| 5 | 74 % | about 118 nmol/L |
| 8 | 46 % | about 74 nmol/L |
| 10 | 34 % | about 55 nmol/L |
| 15 | 16 % | about 25 nmol/L |
| 20 | 7 % | about 12 nmol/L |
| 25 | 3 % | about 5 nmol/L |
| 30 | 2 % | about 3 nmol/L |
| 40 | 0 % | about 1 nmol/L |
- humanA mostly inherited level. More than 90% of the Lp(a) level is determined by variations in the LPA gene, according to the European consensus (Kronenberg et al., 2022); the American Heart Association gives 70 to more than 90% depending on the study (Reyes-Soffer et al., 2022).
- humanThe repeated pattern. The LPA gene carries from 1 to more than 40 copies of the same pattern, kringle IV type 2. The fewer the copies, the shorter the apo(a) and the higher the Lp(a) level (Schmidt et al., 2016).
- humanProduction rather than clearance. With injected tracers, the liver makes small apo(a) more than twice as fast as large ones, with no difference in clearance rate (Rader et al., 1994).
- animal or labQuality control. In baboon liver cells, large apo(a) stay longer in the reticulum (White et al., 1994); some of it is destroyed there before getting out (Kronenberg et al., 2022).
- humanFrom childhood. The gene is fully active by around age 2 and the adult level is reached by around age 5; it then stays stable, but in women it often rises at menopause (Kronenberg et al., 2022).
- hypothesis or modelModel in the drawing. Each gene supplies 160 nmol/L at 3 copies, divided by 2.7 every 6.5 copies; the share of chains that get out follows the same slope. These values are chosen to show the direction of the effect: other variants of the gene also change the level, and two people with the same copies can have very different levels.
How does Lp(a) damage the arteries?
Wall greatly magnified and simplified; the number of particles shows the direction, not an amount.
The steps in text, and where these numbers come from
| Step | Number | What happens |
|---|---|---|
| 1. An LDL carrying a long chain | 1 + 1 an LDL particle and an apo(a) chain | Lp(a) is an LDL, the particle that carries cholesterol, to which the liver attaches a long protein, apo(a). So it carries the cholesterol of an LDL, plus this chain, whose length varies from person to person. |
| 2. It enters the wall and attaches there | Enters the artery wall, like an LDL | Like LDL, Lp(a) passes between the cells that line the artery. In the wall, its chain attaches: the particle stays there more easily and leaves its cholesterol behind. |
| 3. Its oxidized fats attract white blood cells | > 85% of the oxidized phospholipids detected in plasma | The apo(a) chain carries fats damaged by oxidation, the oxidized phospholipids. In the wall, they attract white blood cells, the monocytes, and keep up the inflammation that makes the plaque grow. |
| 4. It looks like the protein that dissolves clots | 78 to 100% similarity between the pieces of apo(a) | The apo(a) gene closely resembles the gene for plasminogen, the protein that dissolves clots. In the lab, apo(a) takes its place on a clot's mesh without being able to dissolve it. In humans, studies have not found this effect. |
| 5. It hardens the aortic valve | × 2 risk of calcium on the aortic valve | Lp(a) also settles on the aortic valve, the heart's exit door. Its oxidized phospholipids push the valve cells to lay down calcium: the valve hardens and opens less well. |
| 6. Overall: more particles, more deposits | +11% risk of a cardiovascular event for every 50 nmol/L | The more Lp(a) particles there are in the blood, the more of them enter the wall and settle on the valve. So the risk rises steadily with the level, even when LDL is low. |
- humanThe particle. Lp(a) is an LDL whose protein, apoB100, is linked to apo(a) by a disulfide bridge (Kronenberg et al., 2022).
- animal or labIn the wall. In mice, apo(a) attaches to the wall through a site on its chain, the same one that binds it to fibrin: with normal apo(a), fatty deposits in the aorta are 5 times more numerous than with an apo(a) whose site is damaged (Boonmark et al., 1997). In humans, imaging shows that Lp(a) triggers inflammation in the wall (Kronenberg et al., 2022).
- humanOxidized phospholipids. In human plasma, more than 85% of the oxidized phospholipids detected by the E06 antibody are bound to Lp(a) (Bergmark et al., 2008).
- animal or labClot. The pieces of apo(a) are 78 to 100% similar to those of plasminogen (Schmidt et al., 2016). In the lab, apo(a) slows the breakdown of clots; studies in humans have not shown this effect (Kronenberg et al., 2022; Reyes-Soffer et al., 2022).
- humanAortic valve. Each copy of the rs10455872 variant of the LPA gene roughly doubles the risk of calcium on the aortic valve (odds ratio 2.05) and multiplies the risk of valve narrowing by 1.68 (Thanassoulis et al., 2013).
- humanRisk. 460,506 adults in the UK Biobank followed for a median of 11.2 years: the risk of a cardiovascular event rises by about 11% for every 50 nmol/L of Lp(a) (Patel et al., 2021).
How much does the risk rise with Lp(a)?
Drag along the curve. Bars: approximate shape of the distribution (our calculation), not a measurement.
The curve in text, and where these numbers come from
| Lp(a) | Relative risk (compared with 20 nmol/L) | People above (approximation) |
|---|---|---|
| 0 nmol/L | × 0,96 | about 100% |
| 20 nmol/L | × 1,00 | about 49% |
| 50 nmol/L | × 1,06 | about 33% |
| 75 nmol/L | × 1,12 | about 25% |
| 100 nmol/L | × 1,18 | about 22% |
| 125 nmol/L | × 1,25 | about 19% |
| 150 nmol/L | × 1,31 | about 12% |
| 200 nmol/L | × 1,46 | about 7% |
| 250 nmol/L | × 1,62 | about 4% |
| 300 nmol/L | × 1,79 | about 3% |
| 400 nmol/L | × 2,21 | about 1% |
- humanRisk slope. 460,506 adults in the UK Biobank: risk of a cardiovascular event multiplied by 1.11 for every 50 nmol/L of Lp(a), almost linearly, with no threshold (Patel et al., 2021). The curve starts at 20 nmol/L, close to the median; this starting point is our choice.
- hypothesis or modelDistribution. Bars drawn from three reference points: a median of 19.6 nmol/L and 12.2% of people without disease at 150 nmol/L or more in the UK Biobank (Patel et al., 2021), about 1 person in 5 above 50 mg/dL (Nordestgaard et al., 2010; Wilson et al., 2022). The shape between these points is our calculation.
- humanUnits. Lp(a) is measured as a number of particles (nmol/L) or as a mass (mg/dL). The mass depends on the length of the chain, so there is no exact conversion: 2 to 2.5 nmol/L per 1 mg/dL is only an order of magnitude (Kronenberg et al., 2022; Reyes-Soffer et al., 2022).
- humanCited threshold. According to the ACC's summary of the 2026 ACC/AHA guideline, more than 125 nmol/L goes with a 1.4 times higher cardiovascular risk, and more than 250 nmol/L with an at least 2 times higher risk. These figures do not start from the same reference as the curve (20 nmol/L): they cannot be compared directly. It is a reference point: the risk rises without steps.
- humanBy ancestry. Medians in the UK Biobank: 16 nmol/L in people of Chinese ancestry, 19 of European ancestry, 31 of South Asian ancestry and 75 of African ancestry. The risk per 50 nmol/L stays similar from one group to another (Patel et al., 2021).
Why does what lowers LDL barely affect Lp(a)?
Each row shows the average change in LDL (blue) and Lp(a) (gold). Tap a row to see why.
Less saturated fatControlled diet trials (Kronenberg et al., 2022; Myagmarsuren et al., 2025)−7 to −11%+10 to 15%
Eating less saturated fat pushes the liver to make more of the receptors that capture LDL: LDL goes down. Lp(a) is poorly captured by these receptors, and its level depends mainly on how much is made: it does not go down. It even rises by 10 to 15%, especially when saturated fat is replaced with carbohydrates, for reasons that are not well understood.
ExerciseTraining studies (Enkhmaa and Berglund, 2022)little changelittle change
Training acts mainly on the fat burned by the muscles and on triglycerides. It does not touch the LPA gene: depending on the study, Lp(a) does not change, or barely.
StatinsMeta-analyses of placebo-controlled trials (Willeit et al., 2018; Tsimikas et al., 2020)−39%+11%
Statins slow down cholesterol production in the liver, which then makes more LDL receptors: LDL drops sharply. Lp(a) does not follow: it rises by 11% on average compared with placebo, 8.5 to 19.6% depending on the trial. In the lab, atorvastatin increases the activity of the LPA gene in human liver cells.
PCSK9 inhibitor (evolocumab)FOURIER trial (Sabatine et al., 2017; O'Donoghue et al., 2019)−59%−27%
PCSK9 inhibitors prevent the destruction of LDL receptors, which become very numerous. At that level, they also capture some of the Lp(a), and production may drop a little: Lp(a) falls by about a quarter.
Pelacarsen (antisense RNA)Phase 2 trial (Tsimikas et al., 2020; Yeang et al., 2022)−7 to −26%−80%
This drug is a strand of RNA that sticks to the message of the LPA gene in the liver and gets it destroyed: apo(a) is hardly made anymore. It acts on production, where the level is set, the opposite of diet. Measured LDL drops a little, mostly because the measurement also counts the cholesterol carried by Lp(a).
Olpasiran (small interfering RNA)OCEAN(a)-DOSE trial (O'Donoghue et al., 2022)−23 to −25%−97%
Same target, different tool: a small RNA that gets the message of the LPA gene cut. One injection every 3 months is enough to make Lp(a) almost disappear. Measured LDL drops by about a quarter: part of that drop comes from the cholesterol in Lp(a), which is counted in the measurement.
Does lowering Lp(a) protect the heart?
- Lp(a)HORIZON, pelacarsen, 8,323 people who had already had cardiovascular disease and an Lp(a) of at least 70 mg/dL: Lp(a) went down, but the trial did not meet its primary endpoint, fewer cardiovascular events (Novartis, September 4, 2026).
- OCEAN(a)-Outcomes, olpasiran: data collection expected to end on December 29, 2026, with results to follow (ClinicalTrials.gov).
- These trials address a question that is still open: is lowering Lp(a) enough to lower the risk it brings?
The table in text, and where these numbers come from
| What changes | LDL | Lp(a) | Why |
|---|---|---|---|
| Less saturated fat | − 7 à − 11 % | + 10 à 15 % | Eating less saturated fat pushes the liver to make more of the receptors that capture LDL: LDL goes down. Lp(a) is poorly captured by these receptors, and its level depends mainly on how much is made: it does not go down. It even rises by 10 to 15%, especially when saturated fat is replaced with carbohydrates, for reasons that are not well understood. |
| Exercise | little change | little change | Training acts mainly on the fat burned by the muscles and on triglycerides. It does not touch the LPA gene: depending on the study, Lp(a) does not change, or barely. |
| Statins | − 39 % | + 11 % | Statins slow down cholesterol production in the liver, which then makes more LDL receptors: LDL drops sharply. Lp(a) does not follow: it rises by 11% on average compared with placebo, 8.5 to 19.6% depending on the trial. In the lab, atorvastatin increases the activity of the LPA gene in human liver cells. |
| PCSK9 inhibitor (evolocumab) | − 59 % | − 27 % | PCSK9 inhibitors prevent the destruction of LDL receptors, which become very numerous. At that level, they also capture some of the Lp(a), and production may drop a little: Lp(a) falls by about a quarter. |
| Pelacarsen (antisense RNA) | − 7 à − 26 % | − 80 % | This drug is a strand of RNA that sticks to the message of the LPA gene in the liver and gets it destroyed: apo(a) is hardly made anymore. It acts on production, where the level is set, the opposite of diet. Measured LDL drops a little, mostly because the measurement also counts the cholesterol carried by Lp(a). |
| Olpasiran (small interfering RNA) | − 23 à − 25 % | − 97 % | Same target, different tool: a small RNA that gets the message of the LPA gene cut. One injection every 3 months is enough to make Lp(a) almost disappear. Measured LDL drops by about a quarter: part of that drop comes from the cholesterol in Lp(a), which is counted in the measurement. |
- humanDiet. Replacing saturated fat with carbohydrates or unsaturated fat raises Lp(a) by about 10 to 15% (Kronenberg et al., 2022). In a meta-analysis of trials, the rise comes mainly from replacement with carbohydrates (Riley et al., 2024). In the DELTA 1 and 2 trials, 175 people fed by the study: LDL −7 to −11%, Lp(a) +15 to 26% on average depending on the diet (Myagmarsuren et al., 2025). Exercise does not change it, or only a little (Enkhmaa and Berglund, 2022).
- humanStatins. Meta-analysis of 6 trials, 5,256 patients: Lp(a) +11% on average compared with placebo, +8.5 to +19.6% in the statin groups depending on the trial; in human liver cells, atorvastatin increases the activity of the LPA gene (Tsimikas et al., 2020). Another meta-analysis, 7 trials and 29,069 patients: LDL −39% on average, with no net change in Lp(a) (Willeit et al., 2018).
- humanPCSK9 inhibitors. Evolocumab: Lp(a) −26.9% (median) at 48 weeks (O'Donoghue et al., 2019); LDL −59% at 48 weeks in the FOURIER trial, from 92 to 30 mg/dL (median) (Sabatine et al., 2017). Alirocumab: Lp(a) −5 mg/dL (median), from a starting level of 21.2 mg/dL (Bittner et al., 2020).
- humanRNA drugs. Olpasiran at 36 weeks, compared with placebo: −70.5% at 10 mg, −97.4% at 75 mg, about −100% at 225 mg (O'Donoghue et al., 2022). Pelacarsen 20 mg per week: Lp(a) −80%, versus −6% on placebo (Tsimikas et al., 2020).
- humanLDL bar. LDL measured in the lab also counts the cholesterol carried by Lp(a). With pelacarsen, it drops by 7 to 26% depending on the dose; with that cholesterol taken out of the calculation, the drop is only 2 to 19% (Yeang et al., 2022). With olpasiran: −22.6 to −24.8% compared with placebo (O'Donoghue et al., 2022). The bar shows the middle of each range. Exercise has no number: its effect on Lp(a) ranges from none to moderate depending on the study.
Frequently asked questions about Lp(a)
Why is Lp(a) mostly inherited?
Because its level depends mainly on the LPA gene received from each parent, and in particular on the number of copies of a repeated pattern in this gene. This number, from 1 to more than 40, is fixed at birth: it sets how much apo(a) the liver can make and let out (Kronenberg et al., 2022; Schmidt et al., 2016).
Why does the Lp(a) level change so little over a lifetime?
Because the production rate, set by the gene, hardly changes. The gene is fully active by around age 2, the adult level is reached by around age 5, and then it stays about the same; in women, it often rises at menopause (Kronenberg et al., 2022).
Why is Lp(a) measured in nmol/L or in mg/dL?
nmol/L counts the particles; mg/dL weighs their mass, which depends on the length of the apo(a) chain. Two people with the same number of particles can therefore have different mg/dL values: there is no exact conversion, only an order of magnitude of 2 to 2.5 nmol/L per 1 mg/dL (Kronenberg et al., 2022).
Why doesn't a low LDL erase the risk from Lp(a)?
Because Lp(a) does not act only through its cholesterol: its chain helps it stay in the wall, it carries oxidized fats that keep up inflammation, and it settles on the aortic valve. The European consensus notes that its risk remains even with a very low LDL (Kronenberg et al., 2022). In 29,069 patients from statin trials, an Lp(a) of at least 50 mg/dL was still linked to a 1.43 times higher risk on statins, compared with less than 15 mg/dL (Willeit et al., 2018).
Why do RNA drugs lower it so much?
Because they target production, where the level is set: they destroy the message of the LPA gene in the liver before apo(a) is made. Lp(a) then drops by 80% with pelacarsen at its highest dose (Tsimikas et al., 2020) and by more than 95% with olpasiran from 75 mg up (O'Donoghue et al., 2022).
See also
Where do the fats we eat go? Egg white vs yolk protein: do you need to cook them? What does a blood pressure of 120/80 mean?
Sources
- Kronenberg F et al. Lipoprotein(a) in atherosclerotic cardiovascular disease and aortic stenosis: a European Atherosclerosis Society consensus statement. European Heart Journal, 2022. doi:10.1093/eurheartj/ehac361
- Schmidt K et al. Structure, function, and genetics of lipoprotein (a). Journal of Lipid Research, 2016. doi:10.1194/jlr.R067314
- Patel AP et al. Lp(a) (Lipoprotein[a]) Concentrations and Incident Atherosclerotic Cardiovascular Disease: New Insights From a Large National Biobank. Arteriosclerosis, Thrombosis, and Vascular Biology, 2021. doi:10.1161/ATVBAHA.120.315291
- Thanassoulis G et al. Genetic associations with valvular calcification and aortic stenosis. New England Journal of Medicine, 2013. doi:10.1056/NEJMoa1109034
- Bergmark C et al. A novel function of lipoprotein [a] as a preferential carrier of oxidized phospholipids in human plasma. Journal of Lipid Research, 2008. doi:10.1194/jlr.M800174-JLR200
- Reyes-Soffer G et al. Lipoprotein(a): A Genetically Determined, Causal, and Prevalent Risk Factor for Atherosclerotic Cardiovascular Disease: A Scientific Statement From the American Heart Association. Arteriosclerosis, Thrombosis, and Vascular Biology, 2022. doi:10.1161/ATV.0000000000000147
- Rader DJ et al. The inverse association of plasma lipoprotein(a) concentrations with apolipoprotein(a) isoform size is not due to differences in Lp(a) catabolism but to differences in production rate. Journal of Clinical Investigation, 1994. doi:10.1172/JCI117292
- White AL et al. Apolipoprotein(a) isoforms: size-dependent retention in the endoplasmic reticulum. Journal of Biological Chemistry, 1994. PubMed 8132643
- Nordestgaard BG et al. Lipoprotein(a) as a cardiovascular risk factor: current status. European Heart Journal, 2010. doi:10.1093/eurheartj/ehq386
- Wilson DP et al. Use of Lipoprotein(a) in clinical practice: A biomarker whose time has come. A scientific statement from the National Lipid Association. Journal of Clinical Lipidology, 2022. doi:10.1016/j.jacl.2022.08.007
- Boonmark NO et al. Modification of apolipoprotein(a) lysine binding site reduces atherosclerosis in transgenic mice. Journal of Clinical Investigation, 1997. doi:10.1172/JCI119565
- Myagmarsuren M et al. Lipoprotein(a) Lipidome: Responses to Reduced Dietary Saturated Fat Intake in Two Randomized Controlled Feeding Trials. Nutrients, 2025. doi:10.3390/nu17193113
- Willeit P et al. Baseline and on-statin treatment lipoprotein(a) levels for prediction of cardiovascular events: individual patient-data meta-analysis of statin outcome trials. The Lancet, 2018. doi:10.1016/S0140-6736(18)31652-0
- Tsimikas S et al. Statin therapy increases lipoprotein(a) levels. European Heart Journal, 2020. doi:10.1093/eurheartj/ehz310
- Sabatine MS et al. Evolocumab and Clinical Outcomes in Patients with Cardiovascular Disease. New England Journal of Medicine, 2017. doi:10.1056/NEJMoa1615664
- O'Donoghue ML et al. Lipoprotein(a), PCSK9 Inhibition, and Cardiovascular Risk. Circulation, 2019. doi:10.1161/CIRCULATIONAHA.118.037184
- Bittner VA et al. Effect of Alirocumab on Lipoprotein(a) and Cardiovascular Risk After Acute Coronary Syndrome. Journal of the American College of Cardiology, 2020. doi:10.1016/j.jacc.2019.10.057
- Tsimikas S et al. Lipoprotein(a) Reduction in Persons with Cardiovascular Disease. New England Journal of Medicine, 2020. doi:10.1056/NEJMoa1905239
- Yeang C et al. Effect of Pelacarsen on Lipoprotein(a) Cholesterol and Corrected Low-Density Lipoprotein Cholesterol. Journal of the American College of Cardiology, 2022. doi:10.1016/j.jacc.2021.12.032
- O'Donoghue ML et al. Small Interfering RNA to Reduce Lipoprotein(a) in Cardiovascular Disease. New England Journal of Medicine, 2022. doi:10.1056/NEJMoa2211023
- Enkhmaa B, Berglund L. Non-genetic influences on lipoprotein(a) concentrations. Atherosclerosis, 2022. PMC9549811
- Riley TM et al. Effects of saturated fatty acid consumption on lipoprotein(a): a systematic review and meta-analysis of randomized controlled trials. American Journal of Clinical Nutrition, 2024. doi:10.1016/j.ajcnut.2024.06.019
- Novartis. Novartis announces Lp(a)HORIZON Phase III topline results for pelacarsen, 2026. novartis.com
- ClinicalTrials.gov. Olpasiran Trials of Cardiovascular Events and Lipoprotein(a) Reduction (OCEAN(a)-Outcomes), NCT05581303. NCT05581303
- ACC/AHA. Guideline on the management of dyslipidemia, Circulation, 2026. doi:10.1161/CIR.0000000000001423
- American College of Cardiology. Lower Sooner: How the 2026 Dyslipidemia Guideline Changes Practice, July 1, 2026. acc.org
True or false?
A low LDL erases the risk from a high Lp(a).
False. Lp(a) also acts through its apo(a) chain and its oxidized fats; its risk remains visible when LDL is low: on statins, an Lp(a) of at least 50 mg/dL is still linked to a 1.43 times higher risk (Kronenberg et al., 2022; Willeit et al., 2018).
An adult's Lp(a) level is largely set by age 5.
True. It depends mainly on the LPA gene; the adult level is reached by around age 5 and changes little afterward, except for a frequent rise at menopause in women (Kronenberg et al., 2022).
Eating less saturated fat lowers Lp(a).
False. These diets lower LDL, but Lp(a) does not go down: it even rises by about 10 to 15% (Kronenberg et al., 2022; Riley et al., 2024).
Cite this page
To cite this page in a paper, an article or a presentation:
Metabolive. "Why doesn't Lp(a) go down with diet?" Updated November 2, 2026. https://metabolive.com/en/everyday/lipoprotein-a.