Highlight
• A pooled analysis of 539,478 participants from two Danish cohorts and UK Biobank found no consistent association between plasma lipoprotein(a) concentrations and risk of Alzheimer’s disease or vascular-related dementia when assessed on continuous scales.
• Accounting for competing risk of death produced divergent absolute risk trends: higher Lp(a) associated with greater absolute risk of vascular dementia in UK Biobank but not in the Copenhagen studies.
• Genetic data (LPA KIV-2 repeat number) from the Copenhagen cohorts showed that very small apo(a) isoforms (≤5th percentile) were associated with a 25% higher subdistribution hazard for Alzheimer’s disease versus larger isoforms (>50th percentile) (sHR 1.25; 95% CI 1.06–1.46).
Background
Dementia is a major global cause of disability and mortality. Alzheimer’s disease (AD) and vascular dementia (VaD) account for most cases, and mixed pathology is common. Cardiovascular risk factors — hypertension, diabetes, dyslipidemia, and smoking — contribute to cerebrovascular disease and may influence dementia risk. Lipoprotein(a) [Lp(a)] is a genetically determined lipoprotein particle containing apolipoprotein(a) (apo[a]) covalently bound to apolipoprotein B. Elevated Lp(a) is a well-established causal risk factor for atherosclerotic cardiovascular disease (ASCVD) and calcific aortic valve disease. Biological features of Lp(a) — atherogenicity, prothrombotic properties, and carriage of oxidized phospholipids — provide plausible mechanisms by which Lp(a) might influence cerebrovascular disease and, indirectly or directly, dementia risk.
Study design
The present analysis pooled data from three large prospective cohort resources: the Copenhagen General Population Study, the Copenhagen City Heart Study (both Denmark), and the UK Biobank. Lipoprotein(a) measurements were available for 539,478 individuals who were followed for up to 30.2 years. Dementia outcomes were classified as Alzheimer’s disease and vascular-related dementia. Genetic data on LPA kringle IV type 2 (KIV-2) repeat number — the major determinant of apo(a) isoform size and a strong inverse correlate of plasma Lp(a) — were available for 117,029 participants from the Copenhagen cohorts.
Analytic methods accounted for competing risk of death (a critical consideration for dementia outcomes studied in older populations, because a higher mortality rate among persons with elevated Lp(a) could bias conventional survival analyses). The investigators examined associations on continuous scales, assessed absolute risks across Lp(a) strata, and used genetic (isoform-size) analyses to probe potential causal relationships.
Key findings
Primary results (pooled observational data): On continuous scales, baseline plasma Lp(a) concentrations were not associated with risk of Alzheimer’s disease or vascular-related dementia in conventional analyses.
Competing-risk analyses (absolute risk): When death was modeled as a competing event, results diverged by cohort source. In UK Biobank (n = 452,989; vascular dementia events = 5,132), absolute risk of vascular-related dementia increased with higher Lp(a) concentrations (trend P = .01). In contrast, in the Copenhagen studies (n = 80,313; vascular dementia events = 2,734), there was no statistically significant trend (P = .42). These discrepant cohort-level findings indicate potential cohort differences in competing risks, follow-up duration, event ascertainment, baseline risk profiles, or residual confounding.
Genetic analyses (LPA KIV-2 repeats): In the Copenhagen cohorts, participants with very small apo(a) isoforms (KIV-2 repeat number ≤5th percentile) had a higher subdistribution hazard for Alzheimer’s disease compared with those with larger isoforms (>50th percentile): sHR 1.25 (95% CI 1.06–1.46). This result suggests that genetically determined small apo(a) isoform size — which correlates with very high Lp(a) levels — may be associated with increased Alzheimer’s risk.
Interpretation and effect sizes: The associations are modest and were not uniformly observed across cohorts or analytical approaches. Continuous Lp(a) levels did not show consistent associations with either AD or VaD. The genetic signal for small apo(a) isoforms and AD (sHR ≈1.25) is suggestive but not definitive for causation; it points to a possible role of extremes of the Lp(a) distribution or isoform-specific biology.
Clinical and statistical considerations
Competing risk of death is a crucial analytical factor in studies of late-life disease. Elevated Lp(a) confers ASCVD risk and could increase mortality before the age at which dementia typically emerges, biasing hazard ratios downward in standard survival models. Subdistribution hazard models and absolute risk estimation — as used here — mitigate that bias and can reveal different implications for public health and clinical practice.
Expert commentary and biological plausibility
Mechanisms that could link Lp(a) to dementia are biologically plausible but indirect. Lp(a) promotes atherosclerosis and may increase cerebral small-vessel disease, lacunar infarcts, and microinfarcts — pathologies that contribute to vascular cognitive impairment and may exacerbate Alzheimer-type pathology. Lp(a) carries oxidized phospholipids and has proinflammatory and prothrombotic properties that could compromise cerebral microcirculation. Conversely, classical Alzheimer’s disease is driven by amyloid-beta accumulation, tau pathology, and neurodegeneration; the role of Lp(a) in these core processes is less clear.
The genetic finding that very small apo(a) isoforms are associated with higher AD risk supports the hypothesis that extremes of Lp(a) biology (isoform-specific effects or very high Lp(a) concentrations) may matter more than modest elevations. However, observational and genetic findings were not wholly consistent across cohorts, indicating the need for replication and mechanistic studies.
Implications for Lp(a)-lowering trials
Several Lp(a)-lowering agents are in clinical development, including antisense oligonucleotides and small interfering RNA targeting apolipoprotein(a). The Phase 3 HORIZON trial of pelacarsen (an antisense oligonucleotide) is an example of a large outcome trial primarily designed to reduce ASCVD events (clinicaltrials.gov NCT04023552). These trial programs offer an opportunity to prospectively collect cognitive outcomes or imaging substudies to determine whether lowering Lp(a) alters trajectories of cognitive decline, incident dementia, or MRI markers of small-vessel disease. Given the current evidence, cognitive endpoints could be considered exploratory or pre-specified secondary outcomes rather than primary endpoints.
Limitations
Key limitations of the pooled analysis include potential heterogeneity across cohorts in Lp(a) assay methods and dementia ascertainment (registries vs clinical adjudication), limited ethnic diversity (primarily Europeans in these cohorts), and residual confounding. Lp(a) is largely genetically determined and stable over life, mitigating concerns about within-person measurement variability, but a single baseline measurement does not capture lifetime exposure dynamics. KIV-2 repeat number is a strong proxy for isoform size but may not capture all functional genetic variation in LPA. Finally, observational associations — even those using genetic proxies — cannot establish causality; randomized controlled trial data will ultimately be needed.
Practical takeaways for clinicians and researchers
• Routine measurement of Lp(a) remains justified for ASCVD risk stratification according to contemporary lipid guidelines. Evidence does not support using Lp(a) to predict dementia risk in general practice at this time.
• Very small apo(a) isoforms or extremely high Lp(a) levels may merit further investigation as potential risk modifiers for Alzheimer’s disease; clinicians should be aware of ongoing research but not change dementia screening or management based on Lp(a) alone.
• Investigators designing Lp(a)-lowering trials should consider embedding cognitive assessments and MRI-based cerebral small-vessel disease endpoints to clarify whether lowering Lp(a) has cerebrovascular or neurodegenerative benefits beyond ASCVD risk reduction.
Conclusion
This large pooled analysis provides reassuring evidence that circulating Lp(a) concentrations, on continuous scales, are not consistently associated with Alzheimer’s disease or vascular-related dementia. However, genetic evidence linking very small apo(a) isoforms to a modestly higher risk of Alzheimer’s disease cannot be dismissed. Combined, the data suggest that common variations in Lp(a) may not materially affect dementia risk, but extremes of Lp(a) biology — high concentrations and small isoform size — warrant further mechanistic, epidemiologic, and interventional study.
Funding and clinicaltrials.gov
See the original publication for detailed funding and conflict-of-interest disclosures (Thomas PE et al., Eur Heart J. 2025). Ongoing Lp(a)-lowering randomized trials include pelacarsen’s HORIZON trial (clinicaltrials.gov NCT04023552), which can provide prospective data on cardiovascular endpoints and amenable platforms for cognitive substudies.
References
1. Thomas PE, Vedel-Krogh S, Nielsen SF, Nordestgaard BG, Frikke-Schmidt R, Kamstrup PR. Lipoprotein(a) and risk of dementia: findings from three cohort studies. Eur Heart J. 2025 Nov 21;46(44):4779-4791. doi: 10.1093/eurheartj/ehaf465. PMID: 40824531.
2. Mach F, Baigent C, Catapano AL, et al.; ESC Scientific Document Group. 2019 ESC/EAS Guidelines for the management of dyslipidaemias: lipid modification to reduce cardiovascular risk. Eur Heart J. 2020;41(1):111-188. doi:10.1093/eurheartj/ehz455.
3. HORIZON trial (pelacarsen; TQJ230) clinicaltrials.gov identifier: NCT04023552.
