Metabolic & Cardiometabolic
PCSK9 inhibitors decreases lipoprotein(a) (~20-30%)
In plain terms: Do PCSK9 inhibitors lower lipoprotein(a), the genetically-driven risk lipid?
Strong support Metabolic & Cardiometabolic 💰 Industry COI noted
Part of: 💊 PCSK9 inhibitors
RefutedContestedStrong support
consensus score 1.00
Yes — evolocumab and alirocumab reduce Lp(a) by roughly 20-30%, a modest but real effect (smaller than for LDL and not enough to fully normalize high Lp(a)).
📅 Last reviewed: 2026-07-14 ⓘ
Evidence ladder
How far up the ladder this claim has climbed. A high consensus on a low rung means "consistent so far," not "proven in people."
Top evidence so far: All trials, pooled (Meta-analysis)
MechanismIn-vitroAnimalObservationalRCTMeta-analysis
How the studies fall
11 support 0 contradict 0 tested null 0 mixed · 11 sources, 10 independent groups
What the evidence shows
Evolocumab and alirocumab lower lipoprotein(a) by roughly 20 to 30 percent, pooled across 27 randomised trials in 11,864 participants. The effect is real but modest, far smaller than their effect on LDL and not enough to normalise a high Lp(a), and two of the trials often cited for it report only LDL outcomes in their text.
The evidence (12)
| Source | Grade | Stance | Quality | Finding |
|---|---|---|---|---|
| Sabatine, Giugliano, Wiviott, Raal, Blom, Robinson, Ballantyne, Somaratne, Legg, Wasserman, Scott, Koren, Stein 2015 · N Engl J Med 372(16):1500-1509 | RCT | supports | low | MAGNITUDE UNVERIFIED. The only lipid outcome in this paper's readable text is LDL cholesterol — 'Patients were followed for a median of 11.1 months with assessment of lipid levels, safety, and (as a prespecified exploratory analysis) adjudicated cardiovascular events', and the sole lipid number reported is the 61% LDL reduction (median 120 -> 48 mg/dL). A case-insensitive search of the full PubMed record returns ZERO occurrences of 'Lp(a)', 'lipoprotein(a)' or 'lipoprotein a'; NEJM 372(16) is paywalled with no PMC deposit, so any Lp(a) result sits in a lipid table this vault cannot read. The previously recorded '~25%' has no provenance in any text the vault holds and has been removed rather than repeated. Stance retained as supports because the direction is not contested by anything readable, but this appraisal is provisional pending full text or the NCT01439880/NCT01854918 results postings. |
| Raal FJ, Stein EA, Dufour R, Turner T, Civeira F, Burgess L, Langslet G, Scott R, Olsson AG, Sullivan D, Hovingh GK, Cariou B, Gouni-Berthold I, Somaratne R, Bridges I, Scott R, Wasserman SM, Gaudet D, RUTHERFORD-2 Investigators 2015 · Lancet | RCT | supports | low | Unverified from the text of record. RUTHERFORD-2's reported endpoints are LDL-C only: 'The coprimary endpoints were percentage change from baseline in LDL cholesterol at week 12 and at the mean of weeks 10 and 12, analysed by intention-to-treat.' No lipoprotein(a) result appears anywhere in the available text (0 hits, case-insensitive, full PubMed record), and this source's provenance is abstract-only, so the previously recorded '~22-32%' Lp(a) reduction has no traceable origin. Stance retained pending full text - the trial plausibly reported Lp(a) as a secondary endpoint - but the magnitude is unconfirmed and the appraisal should be retired if paywalled full text or the NCT01763918 results record cannot substantiate it. |
| Dai H, Zhu Y, Chen Z, Yan R, Liu J, He Z, Zhang L, Zhang F, Yan S 2023 · Endokrynol Pol | meta-analysis | supports | moderate | Systematic review and meta-analysis of RCTs of alirocumab or evolocumab versus placebo in familial hypercholesterolaemia, Embase/MEDLINE/PubMed searched to November 2022: 'Eleven RCTs involved a total of 2408 participants.' The pooled result is a single figure, not a range - 'Alirocumab/evolocumab showed a significant efficacy in reducing Lp(a) [weighted mean difference (WMD): -20.10%, 95% confidence interval (CI): -25.59% to -14.61%] compared with placebo.' That point estimate sits at the very BOTTOM edge of the claim's asserted ~20-30% band and the interval extends well below it (-14.61%), as do both drug subgroups (evolocumab -19.98%, 95% CI -25.23% to -14.73%; alirocumab -20.54%, 95% CI -30.07% to -11.02%) and the 12-week subgroup (-17.61%, 95% CI -23.84% to -11.38%, versus -22.81%, 95% CI -31.56% to -14.07% at >= 24 weeks) - so this row supports the low end of the band, not its centre. The previous extract carried no endpoint value, no interval and no denominator, and its clause about 'extending the effect to a genetically high-Lp(a) population' was not measured: it paraphrases the paper's INTRODUCTION framing that FH is 'characterized by elevated circulating low-density lipoprotein cholesterol (LDL-C) and lipoprotein (a) [Lp(a)] concentrations', while no baseline Lp(a) concentration appears anywhere in the available text. Subgroup denominators are likewise unreported, and the homozygous-FH stratum's interval is very wide ('homozygous FH [HoFH] WMD: -20.04%, 95% CI: -36.31% to -3.77%'), so 'no differential effect' is an undetected difference in an underpowered comparison rather than equivalence. Authors' own hedge: anti-PCSK9 drugs 'may be effective as therapy for reducing serum Lp(a) levels in FH... However, further experimental studies and RCTs are warranted'; the endpoint is a surrogate, with no clinical outcome measured. Independence flagged, not fixed: this is the fourth synthesis voting on this claim, and an 11-RCT FH pool searched to November 2022 necessarily re-passes trials already inside s33298738's 41-trial pool and voting separately as s25282519 (RUTHERFORD-2) and s27793396 (ODYSSEY pooled). Abstract-grade record, so publication-bias/small-study testing, between-study heterogeneity, the identity of the 11 trials, per-subgroup denominators, funding and COI are all the available text does not say. |
| Cao YX, Liu HH, Li S, Li JJ 2019 · Am J Cardiovasc Drugs 19(1):87-97 | meta-analysis | supports | high | Meta-analysis of 27 randomized clinical trials in 11,864 participants of PCSK9 monoclonal antibodies (alirocumab, evolocumab): 'Twenty-seven randomized clinical trials with a total of 11,864 participants were included.' The pooled effect is a single figure, not a range - 'PCSK9-mAbs showed a significant efficacy in reducing Lp(a) (- 21.9%, 95% CI - 24.3 to - 19.5), irrespective of PCSK9-mAb types, treatment duration, participant characteristics, treatment methods, differences of control treatment, baseline Lp(a) levels, and test methods.' That -21.9% sits inside the claim's asserted ~20-30% band, though the interval's lower bound (-19.5%) grazes just outside it. The previously recorded '~21-27%' was a borrowed subgroup denominator: 27% is not a pooled result but the point estimate of one dose stratum - 'The greatest reduction was achieved with 150 mg alirocumab biweekly (- 24.6%, 95% CI - 28.0 to - 21.2) and 140 mg evolocumab monthly (- 26.8%, 95% CI - 31.6 to - 21.9)' - reported under the 27-trial headline. The comparator was also not placebo, as previously stated: the inclusion criterion is 'Randomized clinical trials evaluating the effect of PCSK9-mAbs and control treatment on plasma Lp(a) concentrations were included', and 'differences of control treatment' is itself one of the axes the effect survived, so the control arms were mixed. Searched seven databases plus conference abstracts to July 2018. Meta-regression found 'the more intense low-density lipoprotein cholesterol levels declined during PCSK9-mAb treatment, the greater the reduction in Lp(a) levels'. Authors' own hedge: 'Long-term studies may be needed to confirm the effect of PCSK9-mAbs on Lp(a) in the future.' Abstract-grade record, so publication-bias testing, subgroup denominators, the identity of the 27 trials, funding and COI are all the available text does not say. |
| Mulligan MD, Gandhi RS, Vishwakarma R, Bhattacharya R 2026 · J Clin Lipidol | meta-analysis | supports | moderate | Systematic review and meta-analysis of 31 randomized trials of five PCSK9-targeted agents, random-effects pooling of percent change in Lp(a): 'PCSK9 inhibitors and inclisiran reduced Lp(a) by a pooled mean of -25.76% vs control (95% CI -29.54 to -21.99; P < .0001).' Agent differences came from meta-regression, not head-to-head trials, and were null: alirocumab vs evolocumab +3.3% (-1.40 to 8.07), inclisiran vs evolocumab +4.9% (-2.31 to 12.16). Lerodalcibep and enlicitide showed 'similar approximate 25% reductions', but 'insufficient trial numbers precluded a powered head-to-head comparison'. |
| Bittner VA, Szarek M, Aylward PE, Bhatt DL, Diaz R, Edelberg JM, Fras Z, Goodman SG, Halvorsen S, Hanotin C, Harrington RA, Jukema JW, Loizeau V, Moriarty PM, Moryusef A, Pordy R, Roe MT, Sinnaeve P, Tsimikas S, Vogel R, White HD, Zahger D, Zeiher AM, Steg PG, Schwartz GG, ODYSSEY OUTCOMES Committees and Investigators 2020 · J Am Coll Cardiol | RCT | supports | high | Pre-specified analysis of ODYSSEY OUTCOMES (NCT01663402), the randomised exposure being the claim's own subject: 'One to 12 months after ACS, 18,924 patients on high-intensity statin therapy were randomized to alirocumab or placebo and followed for 2.8 years (median). Lipoprotein(a) was measured at randomization and 4 and 12 months thereafter.' The Lp(a) result is reported in ABSOLUTE units only: 'Alirocumab reduced lipoprotein(a) by 5.0 mg/dl (IQR: 0 to 13.5 mg/dl)', against a baseline of 'median: 21.2 mg/dl; interquartile range [IQR]: 6.7 to 59.6 mg/dl'. The previously recorded '~23%' appears NOWHERE in this paper - it is the ratio of two medians computed by a prior reader, and the median of individual percent changes is not the ratio of medians. Because baseline Lp(a) was heavily skewed (upper quartile above 59.6 mg/dl), a flat 5.0 mg/dl reduction is far below 20% for high-Lp(a) patients, precisely those the claim matters most for; whether the trial's own percent-change figure falls in the claim's asserted ~20-30% band is the paper does not say from the abstract. The dropped dispersion is also part of the finding: the IQR's lower bound is 0 mg/dl, so at or below the 25th percentile there was no Lp(a) reduction at all - this row supports the direction without supporting the effect as reliably delivered per patient. The previous extract's second clause, that Lp(a) reduction 'independently predicted fewer major CV events beyond LDL lowering', is OFF-OBJECT for a claim about lowering Lp(a) and is recorded as a candidate rather than counted here; it is also not a randomised comparison, since alirocumab was randomised but a patient's Lp(a) response was not ('A 1-mg/dl reduction in lipoprotein(a) with alirocumab was associated with a HR of 0.994; 95% CI: 0.990 to 0.999; p = 0.0081' - a model-adjusted, post-randomisation association whose upper bound grazes the null). Sanofi and Regeneron employees are among the authors (noted, not excluded, per this claim's stated policy). Same trial and group_id as s30403574, which holds no row on this claim; these patients probably also vote through the 41-trial pool of s33298738 (flagged, not fixed). Abstract-grade record, so the paper's own percent-change table, the Lp(a) assay method and the funding statement are the available text does not say. |
| Gaudet, Watts, Robinson, Minini, Sasiela, Edelberg, Louie, Raal 2017 · Am J Cardiol 119(1):40-46 | RCT | supports | high | Pooled analysis of the phase 3 ODYSSEY program: 'We evaluated the effect of alirocumab on serum Lp(a) using pooled data from the phase 3 ODYSSEY program: 4,915 patients with hypercholesterolemia from 10 phase 3 studies were included.' Comparators were placebo or ezetimibe; 'Eight studies were conducted on a background of statins, and 2 studies were carried out with no statins.' The result: 'At week 24, reductions from baseline were 23% to 27% with alirocumab 75/150-mg Q2W and 29% with alirocumab 150-mg Q2W (all comparisons p <0.0001 vs controls). Reductions were sustained over 78 to 104 weeks.' Those are two dosing regimens at ONE timepoint, not a continuous 23-29% range, and no magnitude, interval or retention denominator is reported for the 78-104-week timepoints - the previously recorded '~23-29% sustained over >=1.5 years' attached the week-24 figures to the durability window. The independence finding is 'Lp(a) reductions with alirocumab were independent of race, gender, presence of familial hypercholesterolemia, baseline Lp(a), and LDL-C concentrations, or use of statins' - independence from baseline LDL-C CONCENTRATION, not from LDL-C response, which the previous extract asserted. Sponsor-conducted: 4 of 8 authors are Sanofi or Regeneron employees and the biostatistics were done at Sanofi (noted, not excluded, per this claim's stated policy). Abstract-grade record, so the Lp(a) assay method, whether the pooling was prespecified, and per-study denominators are the available text does not say. |
| Farmakis I, Doundoulakis I, Pagiantza A, Zafeiropoulos S, Antza C, Karvounis H, Giannakoulas G 2021 · J Cardiovasc Pharmacol | meta-analysis | supports | high | Systematic review and meta-analysis of double-blind phase 2/3 RCTs of alirocumab or evolocumab vs placebo and/or ezetimibe and/or other lipid-lowering therapy, searched to 17 June 2020, risk of bias by Revised Cochrane RoB 2. The denominator in the previous extract was borrowed: 'Forty-three studies were identified (64,107 patients randomized) and 41 studies were included in the quantitative analysis' - the 64,107 belongs to the 43 identified studies, and the patient total behind the pooled 41 is not stated. The result: 'PCSK9 inhibitors reduced Lp(a) levels by -26.7% (95% CI, -29.5% to -23.9%) with a significant heterogeneity within studies.' That point estimate sits inside the claim's asserted ~20-30% band, but the previous extract's 'robust and consistent effect' is the reader's phrase and is contradicted by the authors' own heterogeneity clause. The conditions the authors attach are part of the finding: 'There was significant difference in Lp(a) change from baseline according to comparator (placebo: mean -27.9%; 95% CI, -31.1% to -24.6% vs. ezetimibe: mean, -22.2%; 95% CI, -27.2% to -17.2%; P = 0.04) and duration of treatment (<=12 weeks: mean, -30.9%; 95% CI, -34.7% to -27.1% vs. >12 weeks: mean, -21.9%; 95% CI, -25.2% to -18.6%; P < 0.01)' - so the durable, beyond-12-week estimate is -21.9%, at the bottom edge of the claimed band, and the headline -26.7% is weighted by short trials. Meta-regression found 'only the mean percentage change from baseline low-density lipoprotein cholesterol due to the intervention is significantly associated with the effect size difference (P < 0.0001)'. Authors' own limit on scope: 'Dedicated randomized controlled trials are needed to establish the benefit of this intervention' - this is a surrogate-endpoint result only. Independence flagged, not fixed: a 41-trial pool searched to mid-2020 necessarily contains the same trials that vote separately on this claim as s30586750 (FOURIER), s31948641 (ODYSSEY OUTCOMES), s27793396 and s24509273, and re-passes the literature of the sibling meta-analysis s30229525. Abstract-grade record, so publication-bias/small-study testing, the identity of the 41 trials, per-subgroup denominators and the funding source are the available text does not say; COI statement is 'The authors report no conflicts of interest.' |
| Qiao W, Feng Y, Wen Z, Dou L, Li Y 2026 · Drugs. 2026 Feb;86(2):217-230 | meta-analysis | supports | high | Umbrella review of 21 published meta-analyses of randomized trials (covering 116 RCTs; 231,796 participants) of PCSK9 inhibitors versus placebo or standard therapy. Lp(a) fell consistently: evolocumab 29.68-46.68%, alirocumab 18.55-26.46%, inclisiran 18.00%, each rated high certainty, and more at higher doses (evolocumab 140 mg 26.8%). The units analysed are previously published meta-analyses, not primary trial data. The authors add that the cardiovascular benefit is 'largely attributable to concomitant LDL-C reduction', with outcome trials targeting Lp(a) still required. |
| O'Donoghue ML, Fazio S, Giugliano RP, Stroes ESG, Kanevsky E, Gouni-Berthold I, Im K, Lira Pineda A, Wasserman SM, Ceska R, Ezhov MV, Jukema JW, Jensen HK, Tokgozoglu SL, Mach F, Huber K, Sever PS, Keech AC, Pedersen TR, Sabatine MS 2019 · Circulation | RCT | supports | high | FOURIER Lp(a) analysis (evolocumab vs placebo, NCT01764633; 'Lp(a) was measured in 25 096 patients in the FOURIER trial'; median follow-up 2.2 years): 'At 48 weeks, evolocumab significantly reduced Lp(a) by a median (interquartile range) of 26.9% (6.2%-46.7%).' The median sits inside the claim's asserted ~20-30% band, but the previously recorded flat 'median 27%' dropped the dispersion, and the lower quartile (6.2%) is far below that band, so this row supports the central estimate without supporting it as a reliably-delivered per-patient effect. Baseline Lp(a) was low in this population: 'The median (interquartile range) baseline Lp(a) concentration was 37 (13-165) nmol/L.' The previous extract's second clause - that higher baseline Lp(a) 'predicted greater CV event reduction' - overstated an OFF-OBJECT finding whose interaction test was not significant: 'Evolocumab reduced the risk of coronary heart disease death, myocardial infarction, or urgent revascularization by 23% (hazard ratio, 0.77; 95% CI, 0.67-0.88) in patients with a baseline Lp(a) >median, and by 7% (hazard ratio, 0.93; 95% CI, 0.80-1.08; P interaction=0.07) in those <=median', and the authors' own wording is that such patients 'tended to derive greater coronary benefit from PCSK9 inhibition'. Lp(a)'s independent prognostic value (placebo arm, adjusted HR quartile 4 vs quartile 1, 1.22; 95% CI, 1.01-1.48) is a different claim and is recorded as a candidate, not counted here. Exposure-randomised test passes: PCSK9 inhibition was itself the randomised variable. Two Amgen employees among the authors (noted, not excluded, per this claim's stated policy); TIMI-group authors Sabatine and Giugliano also appear on sibling rows s25773607 and s24509273 under different group_ids (flagged, not fixed). Abstract-grade record, so whether the Lp(a) analysis was prespecified, the assay method and the funding statement are the available text does not say. |
| Zayed JA, Al-Awamleh NA, Hamad M, Hdaib MIM, Ayesh H 2026 · Diabetes Obes Metab | meta-analysis | supports | low | Network meta-analysis of 51 randomised trials (17,810 participants); the outcome was placebo-adjusted mean difference in percent change in Lp(a). PCSK9-directed agents: evolocumab 140 mg Q2W -31.58 pp (95% CI -38.85 to -24.32), inclisiran 300 mg Q3M -27.17 pp (-40.80 to -13.53); no alirocumab Lp(a) estimate is reported. Dedicated Lp(a) agents were far larger: olpasiran 225 mg -98.94 pp, pelacarsen 80 mg -91.36, muvalaplin 240 mg -85.80, zerlasiran 300 mg -82.71, lepodisiran 400 mg -81.58. Heterogeneity was high (I2 = 90.9%), with funnel-plot asymmetry (Egger p < 0.001) and low certainty. |
| Raal, Giugliano, Sabatine, Koren, Langslet, Bays, Blom, Eriksson, Dent, Wasserman, Huang, Xue, Albizem, Scott, Stein 2014 · J Am Coll Cardiol | RCT | supports | high | Post-hoc pooled analysis of patient-level data from 4 phase II evolocumab trials (n=1,359, 12 weeks), Lp(a) measured by a standardized isoform-independent method: 'Evolocumab treatment for 12 weeks resulted in significant (p < 0.001) mean (95% confidence interval) dose-related reductions in Lp(a) compared to control: 29.5% (23.3% to 35.7%) and 24.5% (20.4% to 28.7%) with 140 mg and 420 mg, dosed every 2 and 4 weeks, respectively, with no plateau of effect.' These are two dosing schedules, not a range; the previously recorded '25-32%' matches neither estimate and 32% exceeds every figure in the paper, and the priority claim 'first systematic demonstration' has no basis in the available text. The authors attach a condition that is part of the finding: 'While the mean percentage of reduction was significantly greater in those patients with baseline Lp(a) of ≤125 nmol/l, the absolute reduction was substantially larger in those with levels >125 nmol/l.' Sponsor-analysed data (6 of 15 authors Amgen employees); abstract-grade record, so subgroup denominators and the identity of the 4 pooled trials are the available text does not say. |
Disagree, or know a study we missed?
We grade by evidence, not opinions. The way to weigh in is to point us to a study we haven't cited (check the evidence table above first), or to flag a problem with one we have. Every submission is reviewed; if it holds up, the grade updates and shows in Science Changes Its Mind.
Opens a short form. You'll sign in with Google so submissions are tied to a real account — we don't display your identity, and we only accept a link we can verify (PubMed, DOI, ClinicalTrials.gov).
Educational only, not medical advice. Grades and scores reflect published evidence weighted by study design and quality; see the methodology.